JP4846038B2 - Analysis apparatus, information processing system, and analysis method - Google Patents

Description

  The present invention relates to an analysis apparatus, an information processing system, and an analysis method, and more particularly to a technique for analyzing a network.

  The problem of global warming has been heard recently. One measure against global warming is to reduce demand for electricity. Electricity is generally supplied by hydropower, solar power, wind power, geothermal power, and other natural power generation, nuclear power generation using nuclear power, and thermal power generation that uses gas, oil, coal, etc. as fuel to generate power. ing. Of these, thermal power generation is a major factor in promoting global warming. Decreasing power demand can reduce the amount of thermal power generation, so global warming can be suppressed accordingly.

  In the fields of IT (Information Technology) and ICT (Information and Communication Technology), the necessity of reducing the power consumed by computers, network devices, and data center devices has begun to be pointed out.

The Green Grid Association, [online], Internet <URL: http://www.thegreengrid.org>

  When measuring the degree of energy saving in a data center (hereinafter referred to as the energy saving level), the amount of work (hereinafter referred to as the amount of work), the energy consumption required to perform the work, and the energy of the entire data center A method for calculating the degree of energy saving from the ratio using consumption is currently used as a standard worldwide. However, the amount of work and the type of work vary depending on the use of the data center and the system or application used, so this method makes it difficult to compare the energy saving level of one data center with the energy saving level of another data center. Moreover, it is difficult to actually measure the work amount of the data center by the work amount measurement method currently proposed.

  The present invention has been made in view of these problems, and an object of the present invention is to provide an analysis technique that can easily measure the amount of work in a node of a network.

  One embodiment of the present invention relates to an analysis apparatus. The analysis device includes an intercept unit that intercepts communication information including node information for identifying a node involved in communication and use information indicating a use of the communication from the network, and the node information and the use from the communication information intercepted by the intercept unit. An extraction unit that extracts information; and a classification unit that classifies communication information based on node information and use information extracted by the extraction unit.

The “node involved in communication” may be a starting point node. Alternatively, it may be an end node.
The “node” may be an information processing apparatus. Alternatively, it may be a server, a router, a switch, or a storage.

  According to this aspect, communication information can be classified according to node information and usage information.

  Another aspect of the present invention is an information processing system. This information processing system includes a network and an analysis device. The analysis device includes an intercept unit that intercepts communication information including node information for identifying a node involved in communication and use information indicating a use of the communication from the network, and node information and use information from the communication information intercepted by the intercept unit. And a classification unit that classifies communication information based on node information and use information extracted by the extraction unit.

  It should be noted that any combination of the above-described constituent elements, or those obtained by replacing the constituent elements and expressions of the present invention with each other between apparatuses, methods, systems, computer programs, recording media storing computer programs, and the like are also included in the present invention. It is effective as an embodiment of

  According to the present invention, it is possible to easily measure the amount of work at a node of a network.

It is the schematic which shows the data center containing the analysis apparatus which concerns on embodiment. It is explanatory drawing for demonstrating the electric power grid | system of the data center of FIG. It is a block diagram which shows the analysis apparatus which concerns on embodiment, and the periphery function and structure. It is a data structure figure which shows an example of the data hold | maintained at the communication information holding part of FIG. FIG. 4 is a data structure diagram illustrating an example of data held in a node information holding unit in FIG. 3. It is a data structure figure which shows an example of the data hold | maintained at the 1st parameter | index holding | maintenance part of FIG. FIG. 4 is a data structure diagram illustrating an example of data held in a coefficient holding unit in FIG. 3. It is a data structure figure which shows an example of the data hold | maintained at the use condition holding | maintenance part of FIG. It is a data structure figure which shows an example of the data hold | maintained at the 2nd parameter | index holding | maintenance part of FIG. FIG. 4 is a data structure diagram illustrating an example of data held in an actual measurement characteristic holding unit in FIG. 3. It is a data structure figure which shows an example of the data hold | maintained at the 3rd parameter | index holding | maintenance part of FIG. It is a representative screen figure of an energy saving examination screen. It is a flowchart which shows a series of processes in the analyzer of FIG. It is a data structure figure which shows an example of the data hold | maintained at the 1st parameter | index holding | maintenance part which concerns on a 1st modification. It is a flowchart which shows a series of processes in the analyzer which concerns on a 2nd modification.

(PUE and DCiE)
The EPA (Environmental Protection Agency) is an energy star that registers energy efficiency indicators for automobiles, home appliances, office automation (OA) equipment, some IT equipment (servers), buildings, etc. ) The energy saving performance of each device is announced through a program called). In the summer of 2010, the operation of the energy star data center, which shows the energy efficiency of the data center, will be started. A method for comparing the work of IT equipment in a data center with energy consumption is the Green Grid Association of a non-profit corporation that conducts data center energy efficiency activities in the United States (see Non-Patent Document 1). Proposed by PUE (Power Usage Effectiveness). PUE is said to be one of the most popular in the world as a numerical value for measuring the energy saving degree of a data center.
However, the mathematical formula for calculating PUE is simple as follows, and the work is not subdivided considering the difference in the use of the data center and the difference in the system and application operated in the data center. .

データセンタのインフラ用のエネルギ利用量は、空調、照明、警備、その他データセンタの建物が利用するすべてのエネルギの利用量を含む。通常このエネルギは電力で賄われる。また、ＩＴ機器の総エネルギも電力からまかなわれる。したがって、これらのエネルギの利用量は電気の利用量すなわち電力量で表され、特にＫＷＨ（キロワット時）の単位で表される。 PUE is defined by Equation 1 below.

Energy usage for data center infrastructure includes air usage, lighting, security, and all other energy usage used by data center buildings. This energy is usually supplied by electricity. In addition, the total energy of IT equipment is also derived from electric power. Accordingly, the amount of energy used is expressed in terms of the amount of electricity used, that is, the amount of power, and in particular in units of KWH (kilowatt hours).

となる。この式２では、５０ＫＷＨ分のＩＴに関する仕事量を作り出すのに、データセンタの総設備は、２００ＫＷＨを使ったことになる。
データセンタのエネルギ総利用量には、ＩＴ機器のエネルギ総利用量も含まれているため、理論的にはＰＵＥが１．０になることはないと考えられる。しかしながら、１．０に近い数値であればあるほど省エネ度が高いデータセンタとされている。本発明者の当業者としての経験から、ＰＵＥは平均的には２程度であり、２を切ると省エネ度がよいと言え、１．５に達すると省エネ度が非常によいと言える。 An example using this PUE will be described. When PUE = 4,

It becomes. In Equation 2, the total equipment of the data center uses 200 KWH to create 50 KWH worth of IT-related work.
Since the total energy usage of the data center includes the total energy usage of the IT equipment, it is theoretically considered that the PUE will not be 1.0. However, the closer the value is to 1.0, the higher the energy saving level. From the experience of the present inventor as a person skilled in the art, it can be said that PUE is about 2 on average, and that the energy saving degree is good when it is less than 2, and the energy saving degree is very good when it reaches 1.5.

As a technique for expressing PUE as a percentage, a technique using DCiE (Data Center Infrastructure Efficiency) (reciprocal of PUE × 100%) has been proposed by Green Grid. DCiE is defined by Equation 3 below.

  The total energy usage of the IT device in Equation 1 that defines PUE and Equation 3 that defines DCiE includes all of the electric energy used when the IT device (particularly the server) is operating. The server consumes a certain amount of power even in an idle state. When the server is in an idle state, the server does not perform a meaningful job, that is, an effective job in view of the use of the server or the data center. That is, when the server is operated in an idle state, it is considered that work is not performed efficiently. If all the servers in the data center are operated in an idle state, the amount of effective work (hereinafter referred to as effective amount of work) can be zero.

  In network connection type (especially Internet connection type) data centers, the trend (trend) of utilization rate (work volume) varies depending on the application system operated there. For example, application server groups that are mainly used to provide Internet search services tend to be used more on holidays than on weekdays, and more frequently on weekdays than in the morning, especially in the evening, even in the evening. is there. As another example, in an Internet stock trading system such as a securities company connected to a stock exchange, the workload increases in response to many transactions being performed during the trading hours of the stock exchange. From the evening when the transaction is closed to the morning, Saturday and Sunday, and public holidays, such a large amount of work does not occur. In the time zone when the stock exchange is closed, it is considered that the broker of the brokerage company only refers to the information of each account, so the workload is less than the trading time zone. In such a state with a small amount of work, the server may be in an idle state. In the idle state, the server does not perform effective work and consumes only power. However, unlike home appliances and the like, data center IT devices cannot simply be turned on and off due to operational requirements, and are energized in some way even if they are in a sleep state. Therefore, power is consumed with a small effective work amount.

  A special-purpose data center having a supercomputer or the like performs effective work non-stop at a server utilization rate of almost 100% until the target program / calculation is completed. Therefore, it can be assumed that almost 100% of the power consumption of the IT device is effectively used (some overhead exists). In this way, there are various uses in the data center, and various uses and characteristics can exist in the systems and applications in the data center, so whether or not effective work is efficiently processed in the above PUE and DCiE. It is difficult to measure. PUE is suitable for time-series comparison of energy saving levels of data centers, but is not suitable for considering whether work is effective or wasteful.

  If 100 servers are operating in an idle state in the data center and the power consumption in these servers is constant at 100 KWH, the data center heated by the heat generated from the 100 servers The power consumed for cooling using the air conditioner is different during the day and at night, and in summer and winter. For example, since the outside air temperature is high in summer, the air conditioning operation increases, and in winter, the outside air temperature is low, so the air conditioning operation decreases. Therefore, PUE differs between summer and winter. Specifically, if the total data center power amount in summer is 200 KWH, the total data center power amount in winter may be about 150 KWH. In this case, PUE is 2.0 in summer and 1. 5 or so. Thus, since PUE fluctuates with changes in the natural environment, it is difficult to say that it is a highly accurate index of energy saving.

According to Moore's Law, the performance of IT equipment will improve as generations progress. It can be said that the server manufactured in 2010 is more efficient in terms of performance than the server manufactured in 2000, and has been improved from the viewpoint of power consumption, and at the same time energy saving. However, in PUE, it is difficult to numerically analyze the difference in energy saving between the 2000 server and the 2010 server. In order to solve this problem, Green Grid and IT industry groups around the world proposed several methods to compare the total effective work of IT equipment with the total energy usage of data centers. However, even with these methods, it is difficult to perform comparative measurement of energy saving levels for data centers in which systems and applications having different uses and purposes are mixed.
In the embodiment, an energy saving degree measurement method is proposed that measures the work efficiency of a data center by monitoring and tabulating the characteristics of the amount of communication on the network.

(DCeP and proxy plan)
So far, several hypotheses have been proposed and methods for measuring effective work have been proposed. However, with the proposed measurement method, it has been found that actual measurement is difficult in consideration of individual operation circumstances of the data center. Green Grid tried to analyze the work more finely than PUE and DCiE, and developed an index called DCeP (Data Center Energy Productivity). DCeP is defined by Equation 4 below.

  For DCeP, the calculation of effective work is theoretically possible, but a calculation application that calculates this effective work can put a considerable load on the server. Therefore, when such a calculation application is run simultaneously in a server that is actually performing an effective work, the calculation application may use excessive memory and CPU resources. By calculating the effective work load, the load on the server increases, and the power consumption of the server itself increases accordingly. As a result, heat is generated and eventually the temperature in the data center rises, increasing the load on the air conditioner. . When DCeP is measured with about one server as a benchmark, such load and heat generation may not be so large, but in a large data center, the number of servers that measure DCeP is from 10 to 100, Every time the number is increased to 200, the power consumption and the air conditioning load accompanying the measurement increase.

Green Grid has proposed the following eight proxies as an effective workload measurement method instead of using DCeP. However, each of these proposals has problems.
Proxy plan Useful Work Self-Assessment and Reporting (Effective Work Self-Assessment and Reporting)
1. Proxy plan Productivity Link Method by Subsetting “Energy Productivity” (DCeP Subset by Productivity Link DCeP)
2. Proxy plan DCeP “Energy Productivity” Subsetting Method by Sampling Workload (DCeP Subset by Sample Workload)
Proxy plan4. Enter / exit data center (total number of communication bits) / (kilowatt hours) (Bits per Kilowatt-hour)
4. Proxy plan CPU load measurement method using SPECint_rate method (Weighted CPU Utilization-SPECint_rate)
Proxy plan6. CPU load measurement method using SPECpower method (Weighted CPU Utilization-SPECpower)
Proxy plan7. Compute Units Per Second (CUPS)
Proxy plan8. Basic OS software workload efficiency method (Operating System Workload Efficiency)

  As another method, the Japanese Green IT Promotion Council (GIPC) has proposed DPPE (Datacenter Performance per Energy) as a data center energy efficiency index. Companies in the United States have proposed a technique using CADE (Corporate Average Datacenter Efficiency). The Australian Computer Association and the British Computer Industry Association, BCS (British Computer Society), are also proposing other measurement methods.

(Problems of conventional methods)
There are two major measurement methods that have been proposed in the world and used in part for testing. The first group is PUE and DCiE advocated by the US Green Grid. DCiE is simply the reciprocal number of PUE expressed in%, and since the numerical value can be compared with the value of%, it was created for simple comparison. That is, DCiE is another notation of PUE, and the content to be calculated is exactly the same as PUE. The PUE is simply a numerical value obtained by dividing the total power consumption of the data center by the power consumption used by the IT equipment in the data center during the same period as shown in Equation 1, so what was the IT equipment doing during that period? Is not considered.

  Among IT devices, for example, a server is generally considered to consume about 30% of electric power during full operation even in an idle state where no effective work is performed. In the latest blade server, power consumption increases even in an idle state because of its high-density circuit configuration and many semiconductor elements installed in a narrow space. Some blade servers consume as much as 70% of the power during full operation in an idle state. There is also data indicating that 30% to 60% of all servers in one data center are in an idle state.

  On the other hand, in the case of a data center that has only a special server group such as a supercomputer, when the calculation for which the supercomputer server group is intended is performed, almost all servers operate at 100% full operation. It is said that it is in full operation without being idle until all calculations are completed. For example, an earth simulator (supercomputer) in Japan will be a data center that continues full operation for a long time during calculation. Further, as a feature of this supercomputer type data center, there is a case where information communication with outside the data center (for example, data transfer to the Internet) is hardly performed while calculation for a special purpose is executed.

  In consideration of the above, PUE simply compares the IT equipment and facility equipment (air conditioner, lighting, etc.) that supports its operation at a certain data center with respect to time. It can only be an indicator. In a data center specialized in securities business, due to its operational characteristics, almost all servers can be in an idle state from midnight when night processing business is completed until the next morning when the exchange is open or on holidays. Therefore, the power consumption of the IT device is lower than that at the time of full operation, so that the PUE is improved (decreased), but the effective work amount is almost zero and nothing is done. That is, a state where the PUE is high but the effective work amount is small may occur. In this state, the degree of energy saving is bad.

With the technique using PUE, it is difficult to measure the actual effective work amount, and therefore it is difficult to measure the power consumed by the IT device to perform effective work. Therefore, Green Grid proposed the above DCeP. In the method using DCeP, as described above, an effective work is defined by the owner of the system or application, and is compared with the power consumption based on the effectiveness. In DCeP, equations and variables are defined as follows.

  In the current data center, a plurality of systems having many different uses are operating in the server group as defined in the definition of Vi, which is one of DCeP variables. Examples of servers having various uses include a mail server that processes electronic mail, a WEB server that provides a WEB page, a DB server that provides a database, and a video streaming server that distributes video images. Servers having these different uses generate different operations and server loads, and the appearance states of operation states such as an idle state, a low load state, a medium load state, and a large load state on the time axis are also different. Moreover, the power consumption in an idle state changes with servers. In addition, power consumption (average value or maximum value) for moving a specific system or application varies depending on the server. Just because 100% is used on the application side, the load on the server is not necessarily 100%. Depending on the performance of the processor (CPU, MPU) of the server, the memory size, the size of the data transfer bus in the server, and the like, a specific application may actually use only 60% of the maximum output that the server can provide.

  Certainly, in the method using DCeP, if it is possible to grasp all the systems and all the associated applications and give a weight to the effective work load of those applications, it is theoretically effective based on the sum of them. The specific gravity of the work volume can be measured. In reality, however, it is difficult to derive all systems and all application usages involved in a general data center, and therefore it is difficult to obtain an effective work ratio. Therefore, Green Grid proposed the above eight proxy plans.

The above eight proxy plans can be roughly classified into the following three types.
(1) Proxy plans 1 to 3
(1-1) Based on the effective work amount (Wi).
(1-2) Measure numerical values obtained from performance monitor.
(1-3) A method of measuring the amount of work using software.
(2) Proxy plan 4
(2-1) Measure the traffic amount of the outward router exiting from the data center.
(3) Proxy plans 5-8
(3-1) Compare the specifications of the server, etc. in the measurement of the degree of energy saving.
(3-2) The proxy plans 5 and 6 are based on the benchmark figures provided by the SPEC organization based on the published benchmark.

  In the proxy plans 1 to 3 belonging to (1) of the three categories, in order to monitor the effective work amount, a performance monitor that is a function of performance measurement already installed in the basic OS is used, or special measurement software is used. Software must be installed. The OS measurement function and measurement software itself may generate new work for the system and increase the load on the server. That is, the idea of causing the server to measure the amount of effective work is unavoidable and a new load accompanying the measurement is unavoidable, which may hinder energy saving.

  The proxy plan 4 belonging to the three classifications (2) simply measures the number of bits of the network output that goes out from the data center, and its mechanism is simple but its use is limited. For example, if the information-providing WEB service data center simply provides WEB pages to external Internet users, the proxy plan 4 can also be used for measurement. However, when the proxy plan 4 is applied to a data center with many systems that perform complicated searches and calculations in the data center and return only the results to the outside, the effective work amount of the server that is processing internally Is almost ignored. In addition, data centers that distribute video such as movies discharge a large amount of data to the Internet due to their characteristics, but as the server side work, video data stored in storage such as disk It simply transfers the data to the network. Therefore, when the proxy plan 4 is applied to such a data center, a complicated work such as calculation is difficult to occur, and only a large amount of data is output, so a very good number is obtained in terms of effective work amount versus power consumption. . On the other hand, in data centers where calculations, complicated searches, and database processing are mainly performed, data sent to the outside is mainly data with a simple screen showing the processing results. Effective work is ignored, and inefficient figures can appear that are far from the effective work of a real data center.

  The proxy schemes 5 and 6 belonging to the three categories (3) take a benchmark method. This is a method for measuring the performance of each server by operating sample measurement software of C language or C ++ provided by a standardization organization in the United States. In the case of plan 6, special hardware is installed in the server, and the amount of power used by the hardware microcode function (firmware of the hardware outside the OS software itself) is measured and the network is connected to the outside. Report via. In Plan 6, a sample benchmark program written in the JAVA (registered trademark) language that runs on the server side called SSJ (Server Side Java) is run. In any case, since the measurement is based on software provided for benchmarking, there is a high possibility that it will produce results that are completely different from actual application software. Therefore, there is a high possibility that it is merely a standard for performance evaluation and power consumption of a server versus other servers (other models of the same manufacturer, equivalent models of different manufacturers, servers with different configurations, etc.).

  The proxy plan 7 is a method of performing measurement by the historical method from the viewpoint that power consumption and server performance may differ between servers of different generations. Here, the actual application usage, job classification, and effective work amount are not considered. The proxy plan 7 is based on the premise that when the server is also viewed in units of time of about 10 years, the calculation speed is faster at the present time than in 10 years ago, and the power consumption is generally reduced. However, when considering a high-density server such as a blade, for example, the processing capability is certainly higher than that of the previous model, but the generated heat is rather increased and the power consumption may be increased. Even if the proxy plan 7 is used to evaluate the energy saving level of the data center, it is better to use many new, high-performance, energy-saving servers that consume less power than to work with many older servers. Can only be obtained. This is because, in the life span of one data center, how many new equipment was introduced in the period of 10 years ago, 5 years ago, 3 years ago, 2 years ago, 1 year ago, etc. We can only find out how much the power consumption of IT equipment has changed.

  The proxy plan 8 is a method of simply comparing the number of basic OSs operating in one data center. This is a technique that increases efficiency by increasing the number of basic OSs relative to the number of servers, and is derived from a virtualization technology that has recently appeared and moves a plurality of basic OSs under one server. The average data center server group is considered to be idle between 30% and 60% of the operating time. If the server is in an idle state in half of the operation time and is operating without effective work, the number of servers can be reduced by the virtualization technology that consolidates the basic OS. The theory of the proxy plan 8 is that the amount of power to be reduced is also reduced. For example, if two servers having 50% idle time are integrated into one virtualization server, it may be said that one server is reduced and ideally power consumption is reduced to about half. However, realistically, which server is in an idle state at which point depends on the characteristics of the application. For example, when two aggregated basic OSs request a load of 100% at the same time, a load of 200% is applied to one server, and the processing capacity of each basic OS can be halved. In other words, although the number of servers is reduced by half, inefficiencies that require twice as much time to process work occur even though the servers are running at 100% load, and extra power is consumed accordingly. there's a possibility that.

ＧＩＰＣ提唱の手法では、ＩＴＥＵを使用して、実際に一つのデータセンタで使われているＩＴ機器の消費電力を合算した総消費電力とメーカが提供している定格電力を合算した総定格電力とを比較する。そして前者が後者より低い場合には、ＩＴ機器には、更なる電力的な負荷をかける余裕があると想定されるので、仮想化や、アプリケーションの統合によりサーバの総数を減らすことが検討されうる。 The measurement method using DPPE proposed by the Green IT Promotion Council (GIPC) is a method of calculating by adding further measurement values to the method using PUE proposed by the Green Grid. In this method, first, ITEU (IT Equipment Utilization) is defined by Equation 6 below.

In the method proposed by GIPC, using ITEU, the total power consumption that is the sum of the power consumption of IT devices actually used in one data center and the power rating provided by the manufacturer is combined. Compare And if the former is lower than the latter, it is assumed that the IT equipment can afford more power load, so it can be considered to reduce the total number of servers by virtualization or application integration. .

ここでは新たにＩＴ機器を大きく３つ、すなわち「サーバ」、「ストレージ」、「ネットワーク機器」に分類しているが、その能力の総和をとっているので、分子はあくまで（ＩＴ機器の総能力）＝（データセンタのＩＴ機器の総消費電力）である。したがってＩＴＥＥはＩＴＥＵと同様である。 The method proposed by GIPC proposes a numerical value called ITE (IT Equipment Energy Efficiency) defined by the following equation (7).

Here, IT equipment is newly classified into three major categories, namely “server”, “storage”, and “network equipment”. However, since the sum of the capacities is taken, the numerator is (total capacity of IT equipment). ) = (Total power consumption of IT equipment in the data center). Therefore, ITEE is the same as ITEU.

ＧＥＣは、データセンタに熱の再生・再利用やソーラーパネルや風力発電装置などの自然エネルギ装置を導入し、それをグリーンエネルギと称した場合、そのグリーンエネルギとデータセンタの総消費エネルギとの比率を示す。ＧＥＣは、いかに環境に配慮したエネルギをデータセンタで発生させ、また利用するかという主旨で提唱されている。 Further, the method proposed by GIPC proposes a numerical value called GEC (Green Energy Coefficient) defined by the following Equation 8.

When GEC introduces natural energy devices such as heat regeneration and reuse, solar panels and wind power generators to the data center and calls it green energy, the ratio of the green energy to the total energy consumption of the data center Indicates. GEC is advocated for how to generate and use environmentally friendly energy in data centers.

ＤＰＰＥは理論上は計算できる。しかしながら、特にサーバやストレージについては、製造元が提供する数値である定格電力と定格電力が得られるとされる条件の下での実際の機器の消費電力とは異なっている場合が多い。また、公表されている定格数値についても、製造メーカにより数字の変更や切り上げ等があり統一されていない。また、同じ機種でも世代が違って異なるバージョンとなると、例えば異なる電源装置が装備されている場合もあるので定格数値は必ずしも同等ではない。すなわち、定格電力はあくまで標準的な平均値に許容値を足した安全数値である場合が多く、総定格電力はそのデータセンタが使い得る電力であるとは限らない。むしろ、これらの定格電力を全て足した場合、かなり余裕を持った数値が生じ、実際に全ＩＴ機器をフル稼働させ１００％負荷をかけても総定格電力にはいたらない場合が多いと考えられる。
似たような事例に自動車の燃費がある。例えば、米国ＥＰＡが提唱し、提示し続けているＥＰＡマイレージは、特定の車種の車両の一般道走行時の燃費および高速道路走行時の燃費である。このＥＰＡマイレージは必ずしも実燃費性能とは一致しない。このことからも、メーカによる定格電力にはある程度の誤差が含まれていると考えるべきである。したがって、ＤＰＰＥはあくまで理論値であると考えるのが正しいと思われる。 DPPE is defined by Equation 9 below.

DPPE can be calculated theoretically. However, particularly for servers and storages, there are many cases where the rated power, which is a numerical value provided by the manufacturer, differs from the actual power consumption of the device under the condition that the rated power can be obtained. In addition, published rating values are not unified due to changes in numbers and rounding up by manufacturers. Moreover, even if the same model has different versions with different generations, for example, since different power supply devices may be equipped, the rated values are not necessarily the same. That is, the rated power is often a safety value obtained by adding an allowable value to a standard average value, and the total rated power is not always power that can be used by the data center. Rather, when all of these rated powers are added, numerical values with a considerable margin are generated, and it is often considered that the total rated power is not reached even when all IT devices are actually fully operated and 100% loaded. .
A similar example is the fuel consumption of a car. For example, the EPA mileage advocated and continuously presented by the US EPA is the fuel efficiency when traveling on a general road and the fuel efficiency when traveling on an expressway of a vehicle of a specific vehicle type. This EPA mileage does not necessarily match the actual fuel consumption performance. From this, it should be considered that the manufacturer's rated power includes a certain amount of error. Therefore, it seems correct to think that DPPE is a theoretical value.

ＣＡＤＥも理論上は計算できる。しかしながら、上記のグリーン・グリッドのプロキシ案と同様に実計測が難しい。特に、ファシリティ設備系の利用度の厳密なパーセンテージ化は難しく、ファシリティ設備の種類の多さ（例えば、空調機器では、水冷却装置や水循環装置など単に冷気を排出するエアコンユニット以外にも、多くの機器が組み合わされてはじめて稼動する）や受電設備、変電設備、配電設備、照明設備等を合わせると用途の違いにより単純にパーセンテージ化し難い。つまり、ファシリティ設備の比重の違いの細分化が難しい。同様に、ＩＴ機器系の利用度の厳密なパーセンテージ化も難しく、サーバの基本ＯＳやサーバに搭載されるアプリケーションやサーバの用途などの相違により、上記のプロキシ案での区分の難しさの通り、ＣＡＤＥでもＩＴ機器の利用度の比重化による計測は難しい。
また、ＣＡＤＥを使用する手法では、米国のデータセンタの基準であるＴｉｅｒ（ティアー）方式の区分を考慮している。Ｔｉｅｒ方式では、特に二重化、多重化、Ｎ＋方式のバックアップなどの設備基準とＩＴ機器の装備による区分をしており、必ずしもＩＴ機器の省エネ化を目的としていない。Ｔｉｅｒ方式では、ＩＴ機器とそのシステム・アプリケーションが停止せずに動くためには、電力消費が多くなっても例えば多重化を考慮する方式を取っている。この場合は、電力消費の省エネ化は無視してもよいとの考え方も含まれていると考えられる。 The CAD proposed by the US company is defined by Equation 10 below.

CADE can also be calculated theoretically. However, as with the above-mentioned green grid proxy plan, actual measurement is difficult. In particular, it is difficult to make a strict percentage of facility facility usage, and there are many types of facility facilities (for example, air conditioners such as water cooling devices and water circulation devices other than air conditioner units that simply discharge cold air) It is difficult to simply make a percentage due to the difference in the use of power receiving equipment, substation equipment, power distribution equipment, lighting equipment, etc. In other words, it is difficult to subdivide the difference in specific gravity of facility facilities. Similarly, it is difficult to make a precise percentage of the usage of IT equipment, and due to differences in the basic OS of the server, the applications installed in the server, the usage of the server, etc. Even with CADE, it is difficult to measure the specific gravity of the utilization of IT equipment.
Further, in the method using CADE, the division of the Tier system that is the standard of the data center in the United States is considered. In the Tier system, classification is made according to equipment standards and IT equipment equipment such as duplex, multiplexing, N + backup, etc., and it is not necessarily aimed at energy saving of IT equipment. In the Tier system, in order for IT devices and their system applications to operate without stopping, for example, a system that considers multiplexing is used even if power consumption increases. In this case, it is considered that the idea that energy saving in power consumption can be ignored is included.

  The measurement method proposed by Australia's NABERS data center is based on the server benchmark, taking into account the multiple measurement methods proposed by Green Grid (PUE, DCiE, DCeP, proxy proposal) and the method using CADE proposed by the US company. It is based on doing. Therefore, this method does not exceed the sampling range, and considering the actual application and system usage and data center characteristics (WEB base, video base, scientific calculation, financial industry, etc.), the proxy of Green Grid There is no big difference with plans 5 and 6.

  A method proposed by DCSG (Data Center Specialist Group), a sub-working group of BCS in the UK, is considered based on DCiE from Green Grid. In the data center simulator created by DCSG, the facility-related components and IT-related devices of each data center are stored in a database. Then, the entire data center is made into a snapshot, and the DCiE in that state is calculated as a standard. Thereafter, a simulation of how the DCiE fluctuates by changing the configuration of the device is performed. As a result, it is predicted what kind of device configuration is changed in terms of facilities and IT equipment to save energy in the data center. After the prediction, the actual configuration is changed, and actual measurement is performed to see what kind of energy saving effect the changed point has. However, this method considers only the power consumption of the entire IT device as long as it is based on DCiE. Therefore, as pointed out in the description of the PUE, the degree of use of each server and the purpose of the server are not taken into consideration, and there is a possibility that a numerical value that is different from complete energy saving may come out.

  Based on the above, the methods for measuring the degree of energy conservation in the six types of data centers currently proposed in the world (the PUE proposed by Green Grid, the method using DCiE, the proxy method proposed by Green Grid, and the Green IT Promotion Council) The method using the proposed DPPE, the method using the CAD proposed by the US company, the method proposed by the Australian NABERS data center, and the method proposed by the British BCS) are too simple to measure and Issues that do not take into account operational characteristics (the actual state of application operation), or conversely, numerical values that are virtually impossible to measure, or numerical values that can cause errors if included in measurement (manufacturer's ratings) Numerical value), or a new load is generated on the server to perform measurement, and the load is higher than the actual operation. There is a problem that the energy consumption occurs.

  The present invention will be described below based on preferred embodiments with reference to the drawings. The same or equivalent components, members, and processes shown in the drawings are denoted by the same reference numerals, and repeated descriptions are omitted as appropriate.

  FIG. 1 is a schematic diagram showing a data center 2 including an analysis apparatus 100 according to an embodiment. The data center 2 includes a first subnet 4, a second subnet 6, a third subnet 8, a PC terminal 10, a first server 12, a second server 14, a third server 16, and a first router 18. The second router 20, the WAN (Wide Area Network) router 22, the SAN (Storage Area Network) storage 24, the NAS (Network Attached Storage) storage 26, the mainframe 28, the supercomputer 30, and node information. A holding unit 138 and an analysis apparatus 100 are provided.

The first subnet 4 is connected to the PC terminal 10, the first server 12, the second server 14, and a node information holding unit 138 (described later in FIG. 5), and relays packets to and from them. The first subnet 4 is connected to the second subnet 6 and the WAN router 22 via the first router 18.
The second subnet 6 is connected to the mainframe 28 and the SAN storage 24 and relays packets to and from them. The second subnet 6 is connected to the first subnet 4 via the first router 18 and the third subnet 8 via the second router 20, and to the WAN router 22 via the first router 18 or the second router 20. Connected.
The third subnet 8 is connected to the third server 16, the NAS storage 26, and the supercomputer 30, and relays packets to and from them. The third subnet 8 is connected to the second subnet 6 and the WAN router 22 via the second router 20.

  The WAN router 22 is connected to an external network 32 such as the Internet. The WAN router 22 sends a packet from the data center 2 to the outside to the external network 32, receives a packet from the outside to the data center 2, and routes the packet to the first router 18 or the second router 20.

The analysis device 100 is connected to the first subnet 4, the second subnet 6, and the third subnet 8, and intercepts packets flowing in each subnet. “Intercepting” may be acquiring the contents of the packet to be intercepted without annihilating it, for example, acquiring the packet regardless of the destination IP address included in the packet.
The analysis device 100 accumulates the contents of the intercepted packets and sorts them by IP address, thereby deriving a work breakdown at the node corresponding to each IP address.
Here, the data center 2 can be regarded as including a network having nodes such as a server, a PC terminal, a router, a storage, and an IO device (such as a printer). Each node communicates with another node or the outside according to a TCP (Transmission Control Protocol) / IP (Internet Protocol) protocol.

  FIG. 2 is an explanatory diagram for explaining the power system of the data center 2 of FIG. The data center 2 further includes a bus bar type table tap type first PDU (Power Distribution Unit) 34 and a bus bar type table tap type second PDU 36.

The first PDU 34 is connected to an external power supply, and supplies power to each of the PC terminal 10, the first server 12, the second server 14, the first router 18, the SAN storage 24, and the node information holding unit 138. The first PDU 34 has a function of measuring the power supplied for each power supply destination and a function of reporting the measured power value to the analysis apparatus 100 via a network not shown in FIG. These functions are realized using known techniques.
The second PDU 36 is connected to an external power supply, and supplies power to each of the third server 16, the second router 20, the WAN router 22, the NAS storage 26, the main frame 28, and the supercomputer 30. The second PDU 36 has the same function as the first PDU 34.

  The analysis apparatus 100 acquires and accumulates measured values of power supplied to each node from the first PDU 34 and the second PDU 36. The analysis apparatus 100 derives the power consumption for each job at the node from the accumulated measurement value.

  The analysis apparatus 100 according to the present embodiment measures the usage of each server in the data center 2 based on the TCP / IP protocol generally used in server-server communication. In addition, the analysis apparatus 100 uses the function of the second step of Energy Star for Server, which is an index of the energy saving degree of equipment performed by the EPA in the United States. In Energy Star for Server 2 conducted by EPA in the United States, registered and certified devices can provide three reporting functions at the microcode (firmware) level of the hardware itself, basic OS and other measurement software In addition, the processing capability of the application and the original server (CPU / MPU of the main body) is supposed to have a hardware management function that does not affect the processing capability. The three reporting functions are 1) the actual power consumption value of the server, 2) the usage rate of the processing processor such as the main body CPU / MPU of the server, and 3) the actual operating temperature of the server.

  If there is no Energy Star for Server2 function performed by EPA in the United States, the actual power consumption is the power outlet device such as the first PDU 34 or the second PDU 36 that has a power usage measurement function that has already been commercialized. Obtained using. This device has a function capable of reporting the power consumption status of each power outlet via a network by itself. The temperature status of each server is acquired using a device having a function of reading the temperature from a temperature sensor external to the server and reporting the temperature via the network. The processor usage rate measurement of the server is performed by a performance monitor function or special measurement software installed in an existing OS device. However, the individual load on the server that occurs when this function is used is measured, and after actual measurement, the value obtained by subtracting the percentage from the total percentage is used as the measurement value.

  In past data centers, for example, a large computer device called a large host computer used a special communication method to transfer and communicate data with peripheral devices (storage, network devices). In the client-server system that appeared later, a special communication system (for example, an IBM channel IO (input / output) device, bus-type peripheral device communication, RS232 or 422 parallel multi-wire between devices Data is exchanged between the server and peripheral devices by the TCP / IP method without using the communication transfer function by connection).

  Currently, communication between IT devices in many data centers is performed by the TCP / IP method. In a small network, communication between devices is performed by Ethernet (registered trademark) using a HUB device. When the network scale becomes large even in a data center, a router that connects and relays subnets including a plurality of HUB devices, a switch that directly and virtually directly connects individual IT devices at high speed, and the like are used. In addition, for high-speed communication, there are also communication through optical fiber devices and fiber channels of special high-speed data transfer devices using the characteristics of optical fibers, but these are based on the TCP / IP protocol. It is carried out. In other words, many devices (servers, storage, network devices, printers, tapes and other storage devices) in the data center communicate with each other using the TCP / IP protocol, and the network includes HUBs, routers, switches (LANs). Equipment) and wide-area telecommunications routers and switches (WAN equipment).

In the TCP / IP protocol, an individual port called a port is prepared in the standard, and a communication application (hereinafter referred to as a communication application) differs depending on the port. In the TCP / IP protocol, when the server A and the server B communicate with each other, a number called an address is used. When communication is performed from server A to server B, a source address (server A address) and a destination address (server B address) are designated, and a port (port) is designated depending on the purpose of communication. Data transfer and communication confirmation / inquiry are performed using the port specified here.
The communication application may be transmission / reception of electronic mail, FTP (File Transfer Protocol), or WEB browsing.
The address is an ID that identifies the node, and is, for example, an IP address or a MAC address or both. The IP address is individually assigned by the data center network administrator. The MAC address is an ID that is assigned to network connection hardware (for example, a LAN card) of a node and uniquely identifies the hardware.

In communication between the PC device D and the WEB server C, communication is performed based on the IP address and MAC address of the PC device D and the IP address and MAC address of the corresponding WEB server C. At the same time, when data is transferred by communication using the HTTP (HyperText Transfer Protocol) method, communication can be established only when the port number is designated as “80”, which is the port number for this type of communication. In this communication, data including information on communication called IP datagram is exchanged between devices via a network based on the rules of the TCP / IP protocol. An IP datagram is also called a packet, and its contents and description method are defined in the standard of the TCP / IP protocol, and communication is possible for the first time by observing this rule. The packet includes, for example, the start point IP address, end point IP address, protocol number, port number, time stamp, transferred data itself, and information necessary for other communications (error handling, size / length). Etc. are included.
Here, the start point IP address and the end point IP address are node information for specifying a node related to communication, and the port number is use information indicating a communication use related to a packet.

  In the TCP / IP standard, a port for standard communication use is prepared. Such ports are defined as “well-know port numbers” and are specified for each communication application.

(Example of part of TCP / IP Well-known port number)
For the file transfer protocol of FTP / file transfer, FTP 21 / TCP and FTP 21 / UDP are defined for control. FTP-Data 20 / TCP and FTP-Data 20 / UDP are defined for data transfer.
Similarly, 23 / TCP and 23 / UDP are defined for TELNET originally created for the purpose of UNIX (registered trademark) screen communication.
For e-mail, SMTP and POP3 communication are used as standard, 25 / TCP and 25 / UDP are defined for SMTP, and 110 / TCP and 110 / UDP are defined for POP3.

  The TCP / IP protocol itself is roughly divided into an IP protocol, a TCP protocol, a UDP (User Datagram Protocol) protocol, and others. In the case of FTP transfer in the above example, the port number “20” is used for data transfer, and the port number “21” is used for control. At the time of transfer, the TCP or UDP protocol is properly used depending on the application.

  When the port number “23” is designated, it is understood that the packet is used for TELNET. When the port number “25” or the port number “110” is designated, it can be seen that the packet is used for an e-mail system. When the WEB browser communicates with the WEB server, the port with the port number “80” is used as a standard.

  The present inventor confirms the contents of a packet flowing on a network in a data center using a TCP / IP protocol network, so that what device and what other device does what. Recognized that it is possible to know whether or not you communicated for the purpose of. The inventor has also recognized that the amount of transferred data can be measured by the number of packets and the amount of loaded data.

  When a network device with intelligent functions of communication exchange called a router or switch throws a question from the network to that device, what it does or does not do (ie communication information and content) ). This function is realized mainly by using functions installed for special analysis of the network. This act of asking and reporting to routers and switches can put an excessive process burden on the routers and switches, especially if too many questions are thrown, the processing capacity of the routers and switches themselves may be reduced. There is a risk that the efficiency of the entire network may be impaired.

  The analysis apparatus 100 according to the present embodiment is installed in the data center 2 and intercepts packets passing through the network. When the node in the data center 2 performs communication according to the TCP / IP protocol, the analysis apparatus 100 intercepts the communication and records the communication status in the storage medium. After that, the analysis apparatus 100 analyzes the data of the communication status in a specific time period, and derives which node communicated with which node for what purpose, and how much data was transferred. To do.

  When analyzing the effective work amount in a node, information on what the work is is necessary. For example, it can be assumed that a server that only performs communication of ports “23” and “110” of the TCP / IP protocol is a mail server and mainly performs email communication. In addition, it can be assumed that a server that mainly performs WWW communication of the WEB server system such as the port “80” of the TCP / IP protocol has a WEB server function as a main business. Also, a server using ports such as “23”, “110”, and “80” at the same time is a server in which an e-mail function and a WEB server are mixed, or e-mail and It can be assumed that this is the PC terminal of the user who is browsing (browsing) the WEB.

A node in the data center 2 has an IP address and a MAC address.
The network manager of the data center 2 manages the IP address and the node to which the IP address is assigned in association with each other. When the node is a server, the network administrator registers and manages information such as the node type, manufacturer, basic OS type, and owner in the node information holding unit 138 from the viewpoint of asset management. . The analysis apparatus 100 refers to the node information holding unit 138 to acquire the node type and usage, the basic OS type, and the like. The analysis apparatus 100 further analyzes from what is recorded in the communication according to the TCP / IP protocol, what kind of use the node performed, and what data transfer to which location. In this way, the analysis apparatus 100 intercepts the network, records the communication contents, and analyzes the data of the communication contents, thereby analyzing what kind of work is performed and simultaneously transferring the amount of data. Can also grasp.

  This method of intercepting the network in the data center 2 makes it possible to clearly work on each server without installing a special monitoring application in the server or using a performance analysis tool for the basic OS of the server. And understand the purpose. In addition, by using the first PDU 34 and the second PDU 36 that can measure the power supplied to each node, it is possible to know how much power each node consumes in a predetermined period. Furthermore, by using these together, it is possible to measure how much power is consumed for what kind of work.

  However, each server does not necessarily perform only network communication, but sometimes performs access to the storage in the server housing, transaction processing, and computation that does not appear on the external network. In these cases, the analysis apparatus 100 acquires the processor usage rate of the server, and simultaneously acquires the application installed in the server from the node information holding unit 138 of the data center 2, and occupies other than the network communication. The workload is also derived and measured as individual workload.

  In order to differentiate the workload for the network and the internal workload that is not so, the benchmark program is run in the data center 2 in a situation where no network is used and the type of port communication that only requires network communication. Let Thereby, it is possible to grasp the characteristics of the server application in a state other than the actual environment operation. The maximum processor load required during network communication can be assumed based on at least the performance of the network interface. Unlike this conventional benchmark, this benchmark method is used to hypothesize the processor load as a percentage by job type, so power consumption at that time cannot be taken into account.

  FIG. 3 is a block diagram illustrating functions and configurations of the analysis apparatus 100 and its surroundings. Each block shown here can be realized by hardware such as a computer (CPU) (central processing unit) and other elements and mechanical devices, and software can be realized by a computer program or the like. Here, The functional block realized by those cooperation is drawn. Therefore, it is understood by those skilled in the art who have touched this specification that these functional blocks can be realized in various forms by a combination of hardware and software.

  The analysis apparatus 100 includes a network interface unit 102, a communication information holding unit 104, a classification unit 132, a totaling unit 130, a first index holding unit 108, a coefficient holding unit 114, a first analysis unit 116, and a usage A status acquisition unit 110, a usage status storage unit 112, an actual measurement characteristic acquisition unit 144, a second index storage unit 118, an actual measurement characteristic storage unit 150, a second analysis unit 152, a third index storage unit 158, A display control unit 120.

The network interface unit 102 includes an intercepting unit 122, an extracting unit 124, a filter unit 126, and a registration unit 128.
The intercepting unit 122 intercepts packets from the first subnet 4, the second subnet 6, and the third subnet 8. The extraction unit 124 extracts the start point IP address, the end point IP address, and the port number from the packet intercepted by the interception unit 122.

  In the filtering mode in which the filter function is turned on, the filter unit 126 selects only packets including the IP address of the measurement target node based on the start point IP address and the end point IP address extracted by the extraction unit 124. To the registration unit 128. More specifically, the filter unit 126 registers the extraction source packet only when at least one of the start point IP address and the end point IP address extracted by the extraction unit 124 matches the IP address of the measurement target node. To the unit 128.

In particular, the filter unit 126 selects nodes to be measured from a plurality of nodes included in the data center 2 in order. That is, the filter unit 126 switches the IP address of the measurement target node in order among the IP addresses of a plurality of nodes included in the network included in the data center 2.
In the filtering off mode in which the filter function is turned off, the filter unit 126 outputs the packet in which the start point IP address, the end point IP address, and the port number are extracted by the extraction unit 124 to the registration unit 128.

For the packet output by the filter unit 126, the registration unit 128 includes a time related to the packet, a start point IP address extracted from the packet by the extraction unit 124, and an end point IP address extracted from the packet by the extraction unit 124. The communication application corresponding to the port number extracted from the packet by the extraction unit 124, the data amount of the packet, and the network work time are associated and registered in the communication information holding unit 104.
The registration unit 128 may register information on one packet in one entry of the communication information holding unit 104, or may register information on a plurality of packets belonging to one session in one entry.

  The classification unit 132 classifies the packet based on at least one of the start point IP address and the end point IP address extracted by the extraction unit 124 and the port number. Hereinafter, a case will be described in which the classification unit 132 classifies packets based on the start point IP address and the port number. However, even when the classification unit classifies a packet based on the end point IP address and the port number, and when the classification unit classifies the packet based on the start point, the end point IP address, and the port number, the following explanation is valid. It will be apparent to those skilled in the art who have touched the specification.

  The classification unit 132 sorts the data held in the communication information holding unit 104 with respect to the start point IP address. Hereinafter, the start point IP address registered in the communication information holding unit 104 is referred to as a registered IP address. The classification unit 132 classifies the packet for each registered IP address according to the communication use. In particular, the classification unit 132 refers to the node information holding unit 138 and acquires information on the original use that is a predetermined communication use for a node corresponding to the registered IP address (hereinafter referred to as a registered node). The classification unit 132 classifies the packet into a packet corresponding to the original use and a packet corresponding to a non-original use which is a communication use other than the original use. The intended use is, for example, mail communication (transmission / reception) corresponding to port number “25” and port number “110” if the registered node is a mail server, and WEB communication corresponding to port number “80” if the registration node is a WEB server. It is. The non-original use is, for example, a communication use other than the original use, such as FTP transfer corresponding to the port number “20” if the registered node is a mail server.

In the present embodiment, the relationship among node usage, communication usage, and port number is defined as follows.
1. When the intended use of the node is a mail server, a port for mail communication such as TCP / UDP port numbers “25” and “100” is often used. Therefore, the intended use is mail communication (including mail transmission and mail reception), and the corresponding port numbers are “25” and “100”.
2. When the original node is an FTP server for file transfer, the TCP / UDP port number “20” is often used. Therefore, the original use is FTP communication, and the corresponding port number is “20”.
3. When the original node is used for a WEB server, the TCP / UDP port number “80” is often used. Accordingly, the original use is WEB communication, and the corresponding port number is “80”.
4). In addition, the screen operations of UNIX servers such as TELNET are mostly used mainly for system maintenance work, and conversely, the frequency of TELNET TCP / UDP port number “23” is usually low. However, if the port “23” is frequently used, it is possible that the server is performing a lot of maintenance work and the actual operation (effective work) is reduced or hindered.
5). A server that is a generic server and should have a mix of work for multiple purposes, simply asks how much data has been transferred to or from the network.
6). In the case of multicast, the port number is not used, and information indicating the multicast included in the packet is used.

The totaling unit 130 totals packets corresponding to the intended use for at least one registered node. In particular, the totaling unit 130 totals the data amount of the packet corresponding to the original use of the registered node for each period of a predetermined length, that is, the time zone, and totals the data amount of the packet corresponding to the non-original use. For example, consider a case where the registered nodes corresponding to the registered IP address “13.3.3.4” are counted in the time zone “02/15/10 11:10:00 to 11:20:00”. From the node information holding unit 138 in FIG. 5, the registered node is “Mail No. 5”, which is a mail server. Based on the classification result in the classification unit 132, the aggregation unit 130 has the start point IP address “13.3.3.4”, the communication usage “mail transmission” or “mail reception”, and the time zone “02/15/10”. 11: 10: 00 ~ 11: 20: 00 ”is added to the packet data amount to obtain the total result for the intended use. In addition, the aggregation unit 130 has a start point IP address of “13.3.3.4”, a communication usage of neither “mail transmission” nor “mail reception”, and a time of “02/15/10 11:10: The total amount of packets included in “00 to 11:20:00” is added to obtain the total result for the non-original use.
The aggregation unit 130 registers the aggregation result in the first index holding unit 108 (described later in FIG. 6).

  The usage status acquisition unit 110 acquires information about the usage status of the node from a node of the network through the first subnet 4, the second subnet 6, and the third subnet 8 at predetermined time intervals or in real time. The usage status of a node is, for example, power consumption (unit is W), device temperature (unit is ° C.) or CPU usage rate (unit is%) if the node is a server, and data transfer amount if the node is storage. (Unit: MB / second), power consumption, and device temperature, and if the node is a network device, data transfer amount, power consumption, and device temperature.

The usage status acquisition unit 110 acquires system performance monitor data installed in the basic OS of the server.
The usage status acquisition unit 110 acquires, from the first PDU 34 and the second PDU 36, a measured value of power supplied to each node as power consumption.
The usage status acquisition unit 110 includes a commercially available device (a temperature sensor or an optical fiber temperature sensor) that can measure the temperature of the node, and a data collection function (a commercially available software package) that stores and reports the temperature data. Or a server-embedded special hardware package) to obtain the temperature of the device.
Alternatively, the usage status acquisition unit 110 uses the above three reporting functions to acquire power consumption, device temperature, and CPU usage rate from nodes registered and certified by Energy Star for Server 2 of the US EPA. To do.

  The usage status acquisition unit 110 associates the acquired information on the usage status, the time zone corresponding to the time when the usage status is acquired, and the node name / IP address of the acquisition destination in association with each other, as shown in FIG. Register in below. When the usage status acquisition unit 110 acquires information at a predetermined time interval, for example, the time zone is set to a period from the acquired time until the predetermined time interval elapses. Alternatively, when the usage status acquisition unit 110 acquires information in real time, a period having a predetermined length is set as a time zone, and information on the usage status is averaged over the period.

The first analysis unit 116 includes a first calculation unit 134, a second calculation unit 136, and a first synthesis unit 142.
The first calculation unit 134 adds the original use data amount and the non-original use data amount for a server (hereinafter referred to as a registration server) among the registered nodes registered in the first index holding unit 108 and performs a corresponding measurement. A total data amount (for example, GB unit) for communication use in a period is obtained.

  In general, it can be measured how much load is applied to the mail server in order to send a predetermined amount, for example, 1 kB of data as an e-mail. Similarly, for various servers, the relationship between the amount of data for communication use and the resulting CPU usage rate can be measured in advance. The coefficient holding unit 114 (described later in FIG. 7) holds a coefficient indicating the relationship (for example, unit of% · minute / GB). If this coefficient is 1, it means that the server has to operate for 1 minute at a CUP usage rate of 1% in order to process, for example, the data amount for communication use for 1 GB. It is assumed that the difference in communication usage is averaged with this coefficient.

  For the registration server, the first calculation unit 134 divides the total data amount for communication use by the length of the corresponding measurement period, for example, the length in minutes, and the time average data amount for communication use (for example, GB / Min). The first calculation unit 134 acquires a coefficient corresponding to the registration server from the coefficient holding unit 114. The first calculation unit 134 multiplies the registration server by the time average data amount for communication use and a coefficient, and calculates a value (for example, unit of%) obtained as a result of the multiplication in the registration server and the measurement period. It is assumed that the CPU usage rate is estimated (hereinafter referred to as the first estimated value).

For example, paying attention to the node name “Mail No. 5” registered in the first index holding unit 108 in FIG. 6, the first calculation unit 134 determines that the measurement period “02/15/10 11:10:00 to 11:20”. : 00 ”,“ 150 (GB) ”is obtained as the total data amount for communication use. Here, 1 GB or less was discarded. The first calculating unit 134 divides “150 (GB)” by the length of the measurement period “10 (minutes)” to obtain a time average data amount “15 (GB / minute)” for communication use. The first calculation unit 134 refers to the coefficient holding unit 114 in FIG. 7 and obtains a coefficient “4 (% · minute / GB)”. The first calculation unit 134 multiplies the time average data amount “15 (GB / min)” for communication use by the coefficient “4 (% · min / GB)” to obtain “60 (%)”. The first calculation unit 134 relates this value “60 (%)” to the communication during the measurement period “02/15/10 11:10:00 to 11:20:00” in the server with the node name “Mail 5”. It is set as the 1st guess value by a process.
The first calculation unit may obtain a first estimated value by obtaining a load applied to the server for each communication application by using a coefficient measured for each communication application and adding them together.

The second calculation unit 136 searches the usage status holding unit 112 using the server name / IP address and the measurement period of the registered server targeted by the calculation of the first estimated value in the first calculation unit 134 as a key, and corresponds. Power consumption, device temperature, and CPU usage rate are extracted. The second calculation unit 136 subtracts the first estimated value from the CPU usage rate extracted from the usage status holding unit 112 and obtains an estimated value of CPU usage rate (hereinafter referred to as a second estimated value) by processing other than processing related to communication. To do.
Processing other than processing related to communication is, for example, virus scanning by a virus checker or writing of data to a hard disk drive. Or, you may be idle and not working.

  The second calculator 136 divides the extracted power consumption into a ratio between the first estimated value and the second estimated value, and the value corresponding to the first estimated value is an estimated value of power used for processing related to communication. The value corresponding to the second estimated value is the estimated value of the power used for processing that is not. The device temperature is the same in that the difference between the device temperature and the outside air temperature is divided into the ratio between the first estimated value and the second estimated value.

  For example, focusing on the node name “Mail No. 5” registered in the first index holding unit 108 of FIG. 6, the second calculation unit 136 receives the node name “Mail No. 5” from the usage status holding unit 112 shown in FIG. ”And the measurement period“ 02/15/10 11: 10: 00 ~ 11: 20: 00 ”, power consumption“ 60 (W) ”, device temperature“ 40 (° C.) ”, CPU usage rate“ 90 ( %) ”. The second calculation unit 136 subtracts the first estimated value “60 (%)” from the CPU usage rate “90 (%)” to obtain the second estimated value “30 (%)”. The second calculation unit 136 divides the power consumption “60 (W)” into a ratio between the first estimated value “60 (%)” and the second estimated value “30 (%)”, that is, 2: 1, and processes related to communication. The estimated value “40 (W)” of the power used for the process and the estimated value “20 (W)” of the power used for the other processes are obtained.

The first synthesizing unit 142 divides the estimated value of power used for processing related to communication calculated by the second calculating unit 136 into a ratio between the original use data amount and the non-original use data amount, The value corresponding to the amount is assumed to be an estimated value of the power originally used for the purpose.
The first synthesizing unit 142 uses the time and node name as keys, and the data held in the first index holding unit 108, the data held in the usage status holding unit 112, and the calculation result (server) by the second calculation unit 136 The second index holding unit 118 (described later with reference to FIG. 9) is associated with the power consumption derived from the communication process, the rising temperature derived from the communication process, the CPU usage rate derived from the communication process, and the power consumption originally derived from the application. Register with.

The actual measurement characteristic acquisition unit 144 applies an average value and a maximum value of actual measurement to the nodes registered in the usage status holding unit 112 (CPU usage rate when the node is a server, and data storage when the node is a storage and network device). The energy saving score is calculated by an equation using the average value and the maximum value of the measured value of the transfer amount), and the average value and the maximum temperature of the device.
The measured characteristic acquisition unit 144 includes a first representative value calculation unit 146 and an energy saving score calculation unit 148.

  The first representative value calculation unit 146 performs statistical processing on data held in the usage status holding unit 112 at a time interval longer than a predetermined time interval in the usage status acquisition unit 110, for example, a time interval of one month or six months.

  For the server among the nodes registered in the usage status holding unit 112, the first representative value calculation unit 146 has an average CPU usage rate that is a time average of the CPU usage rate, and a CPU usage rate within a period subject to statistical processing. The highest CPU usage rate, the average power consumption that is the time average of power consumption, the maximum power consumption that is the highest value of power consumption within the period subject to statistical processing, and the average that is the time average of device temperature The device temperature and the highest device temperature that is the highest value of the device temperature within the statistical processing target period are calculated.

  For the storage and network devices among the nodes registered in the usage status holding unit 112, the first representative value calculation unit 146 calculates the average data transfer amount that is the time average of the data transfer amount, and the statistical processing target period. Maximum data transfer amount that is the maximum value of data transfer amount, average power consumption that is the average of power consumption over time, maximum power consumption that is the maximum value of power consumption within the period subject to statistical processing, and time average of device temperature And the highest device temperature that is the highest value of the device temperature within the statistical processing target period.

なお、ＣＰＵ使用率の単位は（％）であるがここでは単位は無視し、数値のみを扱う。同様に電力の単位は（Ｗ）、温度の単位は（℃）であるが単位は無視し数値のみを扱う。すなわち、全ての単位を無視する。

なお、データ転送量の単位は（ＭＢ／秒）であるがここでは単位は無視し、数値のみを扱う。同様に電力の単位は（Ｗ）、温度の単位は（℃）であるが単位は無視し数値のみを扱う。すなわち、全ての単位を無視する。

なお、データ転送量の単位は（ＭＢ／秒）であるがここでは単位は無視し、数値のみを扱う。同様に電力の単位は（Ｗ）、温度の単位は（℃）であるが単位は無視し数値のみを扱う。すなわち、全ての単位を無視する。 The energy saving level score calculation unit 148 calculates the server energy saving level score according to the following formula 11, calculates the storage energy saving level score according to the following formula 12, and the network device energy saving level score according to the following formula 13. Is calculated.

The unit of the CPU usage rate is (%), but here the unit is ignored and only the numerical value is handled. Similarly, the unit of power is (W) and the unit of temperature is (° C.), but the unit is ignored and only numerical values are handled. That is, all units are ignored.

The unit of the data transfer amount is (MB / second), but here the unit is ignored and only the numerical value is handled. Similarly, the unit of power is (W) and the unit of temperature is (° C.), but the unit is ignored and only numerical values are handled. That is, all units are ignored.

The unit of the data transfer amount is (MB / second), but here the unit is ignored and only the numerical value is handled. Similarly, the unit of power is (W) and the unit of temperature is (° C.), but the unit is ignored and only numerical values are handled. That is, all units are ignored.

  For the server, the energy saving score calculation unit 148 calculates the node name, type, IP address, owner classification, and value calculated by the first representative value calculation unit 146 (average CPU usage rate, maximum CPU usage rate). , Average power consumption, maximum power consumption, average device temperature, maximum device temperature) and the energy saving score calculated by the energy saving score calculating unit 148 are associated with each other and registered in the measured characteristic holding unit 150 (described later in FIG. 10). To do. For storage and network devices, the energy conservation score calculation unit 148 calculates the node name, type, IP address, owner classification, and value calculated by the first representative value calculation unit 146 (average data transfer amount, maximum (Data transfer amount, average power consumption, maximum power consumption, average device temperature, maximum device temperature) and the energy saving score calculated by the energy saving score calculation unit 148 are associated with each other and registered in the measured characteristic holding unit 150.

The second analysis unit 152 includes a second representative value calculation unit 156 and a second synthesis unit 154.
The second representative value calculation unit 156 statistically processes the data held in the second index holding unit 118 at a time interval similar to the time interval used by the first representative value calculation unit 146.
The second representative value calculation unit 156, for the nodes registered in the second index holding unit 118, the average original use data amount that is the time average of the original use data amount and the average that is the time average of the non-original use data amount Non-original use data amount, average communication-derived power consumption, which is the time average of communication-derived power consumption, average communication-derived increase temperature, which is the time average of communication-derived rising temperature, and average, which is the time average of communication-derived CPU usage rate A communication-derived CPU usage rate and an average original use power consumption that is a time average of the original use power consumption are calculated.

  The second synthesizing unit 154 uses the node name as a key, the data held in the measured characteristic holding unit 150, and the calculation result by the second representative value calculating unit 156 (average original usage data amount, average non-original usage data amount, average The communication-derived power consumption, the average communication-derived rising temperature, the average communication-derived CPU usage rate, and the average original application power consumption) are associated and registered in the third index holding unit 158 (described later in FIG. 11).

  The display control unit 120 displays various parameters for each node based on at least one of the first index holding unit 108, the second index holding unit 118, and the third index holding unit 158 in response to a request from the user. Is displayed on the monitor 140. The display control unit 120 displays, for example, an energy saving examination screen 336 (described later in FIG. 12) on the monitor 140.

FIG. 4 is a data structure diagram showing an example of data held in the communication information holding unit 104. The communication information holding unit 104 holds the time 202, the start point IP address 204, the end point IP address 206, the communication application 208, the data amount 210, and the network work time 212 in association with each other.
The time 202 is a time related to the packet. For example, if the packet includes a time stamp, the time 202 is a time indicated by the time stamp. Alternatively, the time when the packet was intercepted by the intercepting unit 122 may be used.
The start point IP address 204 and the end point IP address 206 are the start point IP address and the end point IP address extracted by the extraction unit 124, respectively.
The communication application 208 is a communication application corresponding to the port number extracted by the extraction unit 124.
The data amount 210 indicates the amount of data loaded in the packet in units of bytes (B).
The network work time 212 indicates a time during which communication processing related to a packet has continued.

FIG. 5 is a data structure diagram showing an example of data held in the node information holding unit 138. The node information holding unit 138 includes the node name / location 214 of the node, the subnet classification 216 to which the node belongs, the node IP address 218, the node device type 220, the node configuration information 222, and the basic OS of the node. The usage 224 and the node owner classification 226 are stored in association with each other.
When the classification unit 132 obtains information on the original use of a node by referring to the node information holding unit 138, the classification unit 132 may obtain the original use by referring to the basic OS / use of the node. The intended use may be determined based on any combination of the entry items of the node included in the holding unit 138.

FIG. 6 is a data structure diagram illustrating an example of data held in the first index holding unit 108. The first index holding unit 108 includes the node name 228 of the registered node, the original use 230 of the registered node, the main port number 232 that is a port number corresponding to the original use, the original use data amount 234, and the non-original use. A data amount 236 and a measurement period 238 corresponding to a period of a predetermined length, that is, a time zone used by the totaling unit 130 are stored in association with each other.
The original use data amount 234 is a data amount obtained as a result of adding the data amount of the packet corresponding to the original use of the registered node within the measurement period.
The non-original use data amount 236 is a data amount obtained as a result of adding the data amount of the packet corresponding to the non-original use of the registered node within the measurement period.

  FIG. 7 is a data structure diagram illustrating an example of data held in the coefficient holding unit 114. The coefficient holding unit 114 holds the server name 240 and the coefficient 242 in association with each other.

  FIG. 8 is a data structure diagram showing an example of data held in the usage status holding unit 112. The usage status holding unit 112 includes the node name / IP address 244 of the node whose usage status is acquired by the usage status acquisition unit 110, the recording time zone 246 corresponding to the time when the usage status is acquired, and the acquired power consumption. 248, the acquired device temperature 250, the acquired CPU usage rate 252, and the acquired data transfer amount 253 are stored in association with each other.

FIG. 9 is a data structure diagram illustrating an example of data held in the second index holding unit 118. The second index holding unit 118 includes a node name 254, an original use 256, a main port number 258, an original use data amount 260, a non-original use data amount 262, a measurement period 264, power consumption 266, The device temperature 268, the CPU usage rate 270, the data transfer amount 271, the communication-derived power consumption 272, the communication-derived rising temperature 274, the communication-derived CPU usage rate 276, and the intended use power consumption 277 are associated with each other. Hold.
The communication-derived power consumption 272 is an estimated value of power used for processing related to communication calculated by the second calculation unit 136. The communication-derived increased temperature 274 is a portion that is predicted to have been increased by processing related to communication among the difference between the device temperature and the outside air temperature calculated by the second calculation unit 136. Here, the outside air temperature is 25 (° C.). The communication-derived CPU usage rate 276 is a first estimated value calculated by the first calculation unit 134.
The intended use power consumption 277 is an estimated value of the power used for the intended use, calculated by the first combining unit 142.

FIG. 10 is a data structure diagram showing an example of data held in the actual measurement characteristic holding unit 150. The actual measurement characteristic holding unit 150 includes a node name 278, a type 280, an IP address 282, an owner classification 284, an average CPU usage rate 286, a maximum CPU usage rate 288, an average data transfer amount 290, and maximum data. The transfer amount 292, the average power consumption 294, the maximum power consumption 296, the average device temperature 298, the maximum device temperature 300, and the energy saving score 302 are stored in association with each other.
Note that the DASD (Direct Access Storage Device) in FIG. 10 is a dedicated disk device used in an old host computer.

  FIG. 11 is a data structure diagram illustrating an example of data held in the third index holding unit 158. The third index holding unit 158 includes a node name 304, an original use 306, an average original use data amount 308, an average non-original use data amount 310, an average communication-derived power consumption 312, and an average communication-derived increase temperature 314. The average communication-derived CPU usage rate 316, the average original application power consumption 317, the average CPU usage rate 318, the maximum CPU usage rate 320, the average data transfer amount 322, the maximum data transfer amount 324, and the average power consumption 326 The maximum power consumption 328, the average device temperature 330, the maximum device temperature 332, and the energy saving score 334 are held in association with each other.

  FIG. 12 is a representative screen diagram of the energy saving review screen 336. The energy saving examination screen 336 has at least one node state display area 338 for displaying the state of the node designated by the user.

  In the above-described embodiment, examples of the holding unit are a hard disk and a memory. Based on the description in this specification, each unit is realized by a CPU (not shown), an installed application program module, a system program module, a memory that temporarily stores the contents of data read from the hard disk, and the like. It is understood by those skilled in the art who have touched this specification that they can do this.

  The operation of the analysis apparatus 100 configured as above will be described. FIG. 13 is a flowchart showing a series of processes in the analysis apparatus 100. The intercepting unit 122 intercepts the packet from the network (S402). The extraction unit 124 extracts the start point address and the port number from the intercepted packet (S404). The registration unit 128 registers the extracted start point address, port number, and packet data amount in the communication information holding unit 104 (S406). The classification unit 132 sorts the data held in the communication information holding unit 104 by the start point address (S408). The classification unit 132 classifies the sorted data into an original use and a non-original use (S410). The tabulation unit 130 tabulates the classified data (S412). The first analysis unit 116 estimates the power used for processing related to communication at the node from the counting result (S414).

  According to analysis apparatus 100 according to the present embodiment, packets are intercepted from the network, and the intercepted packets are classified based on the start point IP address and the port number. Therefore, the communication data amount can be measured for each starting point IP address and each port number. Accordingly, it is possible to trace how much work corresponding to which communication application (e-mail communication, WEB browsing, etc.) has been performed for a desired node to be analyzed from a packet flowing on the network without directly inquiring the node. Since the node is not inquired, there is almost no burden on the node side accompanying the analysis of the workload. Since the analysis apparatus 100 intercepts the network, the analysis apparatus 100 contributes little to the measurement and can be grasped.

  Since the IP address indicates the subject of work, and the port number indicates the purpose of communication, in the present embodiment, the breakdown of the work performed at the node can be derived using them, and particularly the effective work amount can be measured. In this respect, it is greatly different from the conventional technique in which it is practically difficult to measure the effective work amount. According to the present embodiment, by analyzing the amount of work in a node, it is possible to measure how much effective work the node has performed, and a good index for energy saving of the data center 2 can be obtained.

Further, for example, when analyzing the mail server by the analysis apparatus 100, it may be understood that mails from overseas bases are frequently received at night and that in-house mails are frequently transmitted and received during the day. Further, even when the CPU usage rate of the mail server is almost occupied only by sending and receiving in-house mail, it can be understood, so it is possible to take measures such as self-inhibition of in-house mail.
In addition, when a mail server is provided for each department in the company, the tendency of electronic mail can be known for each department. For example, since the analysis by the analysis apparatus 100 can know whether there are many attachment files or not, the departments using a mail server that processes many attachment files are attached files except for design departments that use CAD. It is recommended to make efforts to reduce energy consumption and promote energy saving. Alternatively, it may be an index for considering the introduction of groupware.

Further, since the analysis apparatus 100 knows the start point IP address and the end point IP address of the packet, the communication path may be optimized based on the information. For example, consider a company with offices in Tokyo, New York and San Francisco. In this company, a packet from a Tokyo data center is sent to New York and then routed to San Francisco. If the data center in Tokyo is analyzed using the analysis device 100 and it turns out that there are many packets from Tokyo to San Francisco, this company will change to a route that sends packets from Tokyo to other locations via San Francisco. Thus, the cost can be reduced.
Thus, according to the analysis apparatus 100, the characteristics of each node can be grasped.

  Further, according to the analysis apparatus 100 according to the present embodiment, since the intended use power consumption can be obtained, the power consumed for effective work in the node or a value strongly related to such power is obtained. Can be considered. Therefore, for example, a value obtained by dividing the original use power consumption by the original use data amount is regarded as a ratio between the effective work amount and the power consumption, and the energy saving degree can be determined between the same type of nodes by comparing the values.

  Moreover, in the analysis apparatus 100 according to the present embodiment, the intercepted packet information held in the communication information holding unit 104 is classified into an original use and a non-original use. Therefore, as shown in the first index holding unit 108 in FIG. 6, it is possible to obtain the original use data amount and the non-original use data amount for the desired analysis target node. Thereby, it can be tracked how much the node to be analyzed is used for the original purpose from the ratio of both. This is one of the good indexes for saving energy in the data center 2. For example, the efficiency of the data center can be improved by finding a node that is not used for the original purpose and taking an appropriate measure.

  In the analysis apparatus 100 according to the present embodiment, the second index holding unit 118 has a form in which the first index holding unit 108 and the usage status holding unit 112 are associated with each other. Therefore, when considering the energy saving of the data center 2, it is possible to grasp the workload and usage status in the node in association with each other. In other words, it is possible to analyze how much CPU processing power is used, how much power is consumed, and how much temperature has been increased by what and when the node has done what work.

  For example, by looking at the breakdown of the workload of the server whose CPU usage rate is 100%, it is possible to know how much of the CPU usage rate of 100% is actually used for effective work. If most of the CPU usage rate is not used for effective work, the server should be considered for energy saving. In the conventional technique using PUE or the like, since the work breakdown cannot be seen, such a determination cannot be made. However, the analysis apparatus 100 according to the present embodiment can be used.

  Also, for example, if there is a server with both low power consumption and effective workload, it looks like a server that is good for energy saving because of low power consumption when looking only at the usage status, and no measures are taken against this server. It is likely not. However, in the present embodiment, when the usage status and the workload are presented in a form that can be compared, it can be seen that the effective workload is also low, so in order to improve the efficiency of the data center and promote further energy saving, For example, it is possible to take measures such as turning work from another high-load server by virtualization. Therefore, the contribution to energy saving of the data center is large.

  In addition, for example, when there is a server with a large exhaust heat although there is no problem in the effective work amount, it can be found. Therefore, if such a server exists in the hot aisle, it can be proposed to the server owner that it is better to move it to another location.

Further, in analysis device 100 according to the embodiment, first analysis unit 116 calculates communication-derived power consumption and communication-derived increased temperature from data held in first index holding unit 108. Therefore, it is possible to analyze the power consumption and the temperature of the device at the analysis target node by subdividing the work performed at the node.
For example, if the communication-derived power consumption is significantly smaller than the power consumption that can be seen from the usage status, there is a possibility that, for example, a virus checker was running during that time period. Furthermore, when an event that significantly reduces communication-derived power consumption occurs frequently, it can be determined that there is a possibility that some trouble is associated with the virus checker, and appropriate countermeasures can be taken. Therefore, the efficiency of the data center 2 can be further increased.

  In the analysis apparatus 100 according to the embodiment, the filter unit 126 includes the IP address of the measurement target node based on the start point IP address and the end point IP address extracted by the extraction unit 124 in the filtering mode. Only the packet is selected and output to the registration unit 128. Therefore, the amount of data held in the communication information holding unit 104 can be reduced within a range that does not affect measurement.

  In addition, the filter unit 126 selects nodes to be measured from a plurality of nodes included in the data center 2 in order. In other words, not all nodes are measured at one time, but the nodes to be measured are, for example, nodes connected to the first subnet 4 on Monday, nodes connected to the second subnet 6 on Tuesday, etc. Change in order. As a result, it is possible to reduce all the data held in the communication information holding unit 104 within a range that does not affect the measurement and set all nodes of the network as measurement targets.

  Moreover, in the analysis apparatus 100 according to the present embodiment, an energy saving score is calculated for each node and associated with the work amount. Therefore, the energy saving level of the node can be more easily grasped by the energy saving level score. Also, by comparing the workload and the energy conservation score, for example, for two servers of the same use, the effective workload is not inferior to each other, but one is inferior in terms of energy conservation score than the other In the next replacement of the equipment, it can be determined that the person with the lower energy saving score is replaced with priority.

  Consider a server that is being virtualized. Conventionally, processor utilization rates of individual server images (individual guest OSs) on the parent (host) virtualized OS are mainly obtained and compared. Then, the guest OS is relocated to another virtual host OS server that is vacant in terms of processor utilization. However, in the case where the guest OS is moved between the virtualization servers using only the processor usage rate as a scale in this way, the breakdown of work actually performed by the guest OS is not taken into consideration. Therefore, it cannot be said that the relocation based on the effective work volume is performed.

When analyzing work in a virtualized server using the analysis apparatus 100 according to the present embodiment, since the work amount is known for each virtualization platform, a highly accurate guest OS is relocated based on the effective work amount. Is possible.
In addition, for example, it is possible to know how much the situation has been improved by virtualization. That is, the degree of improvement can be known by comparing the effective work amount and power consumption before and after virtualization. In addition, after virtualization, the current situation is that little attention is paid to the work performed by each guest OS, but the analysis apparatus 100 keeps track of the effective work amount for each guest OS, so that it is not used with time, for example. Guest OS can be specified. Therefore, the performance of other guest OSs can be improved by removing the guest OS thus identified. This contributes to improving the efficiency of the data center.

(Summary)
In the past, system operators have mainly evaluated server performance using server processor usage and memory utilization in system operation, and server integration and application Relocation, distribution, integration, and server update (exchange / replace with new equipment). However, particularly in the recent economic downturn, there is a need to further reduce the operational cost of the entire data center by further effective use of servers and further reduction of power consumption.
Therefore, by using the analysis apparatus 100 according to the present embodiment, it is possible to analyze in detail whether the work in the server is inward (to the server) or outward (from the server), and the work is performed in the data center. You can also analyze whether it is an external job or another server or application in the data center. Therefore, based on this analysis, high-accuracy server integration, distribution, application integration, distribution, equipment update, enhancement, abolition, etc. can be performed for energy saving.

  The configuration and operation of the analysis apparatus according to the embodiment have been described above. These embodiments are exemplifications, and it is understood by those skilled in the art that various modifications can be made to each component and combination of processes, and such modifications are within the scope of the present invention. .

In the embodiment, the classification unit 132 has described the case of classifying the packet into a packet corresponding to the original use and the packet corresponding to the non-original use which is a communication use other than the original use, but is not limited thereto. For example, the classification unit may classify the packets related to one session and classify them for each communication use. Alternatively, the classification unit may classify the packet for each communication use regardless of the original use.
The classification unit according to the first modification classifies the packets registered in the communication information holding unit 104 for each communication use regardless of the original use. The aggregation unit according to the first modification aggregates the classified packets and registers them in the first index holding unit 170 according to the first modification.
FIG. 14 is a data structure diagram illustrating an example of data held in the first index holding unit 170 according to the first modification. The first index holding unit 170 includes a node name 340, an original use 342, a main port number 344, an email communication 346 that is a packet data amount related to email communication, and a packet data amount related to WEB communication. A certain WEB communication 348, an FTP communication 350 that is a data amount of a packet related to FTP communication, a multicast 352 that is a data amount of a packet related to multicast, a data transfer 354 that is a data transfer amount in a storage or a router, and a measurement time 356 ,including.

In the embodiment, the case where the intercepting unit 122 intercepts packets from the first subnet 4, the second subnet 6, and the third subnet 8 has been described, but is not limited thereto. For example,
Consider a situation where a switch is used in the network of the data center 2. If the switch is completed by itself, an analysis device may be installed between the port of the switch and the server connected to the port to intercept the packet. If the switch is connected to another switch or router, an analysis device is installed between the switch port and the external partner port connected to the port to intercept the packet. Also good.

  Alternatively, in recent years, there is also one huge device that can also be used as a switch, HUB, and router that can delegate the switch function of the network in the data center, and a device that can be completed in one centralized network hardware environment. It is commercially available. If such a device is used, packets may be intercepted from the network traffic of the centralized network controller.

  Alternatively, the intercepting unit may be connected to the WAN router 22 and intercept the packet therefrom. This facilitates work analysis in the data center 2 unit. For example, it is possible to determine whether the data center 2 handles a large amount of e-mail, processes a lot of WEB, or specializes in streaming.

  In the embodiment, when a server only performs mathematical calculation processing, the amount of communication between the server and an external device is reduced, or data transfer between a specific storage is mainly performed. In the case of parallel processing, there are cases where the server and the other server specialize in exchanging data and control signals. In this case, when the server is registered in the node information holding unit 138 as being specialized for mathematical calculation and the workload is calculated, the CPU usage rate of the server is more than the analysis of the workload by the network. The amount of work, power consumption, and heat generation may be analyzed.

  In the embodiment, for example, as illustrated in FIG. 13, the case has been described in which data is aggregated (S412) from packet interception (S402) in the analysis apparatus 100, but the present invention is not limited thereto. For example, the analysis apparatus may perform from packet interception to data classification at any time to accumulate classification result data, and may aggregate the classification result data when a predetermined condition is satisfied. Here, the predetermined condition is, for example, that a predetermined period has elapsed since the previous aggregation, or that the amount of classification result data has reached a predetermined amount.

FIG. 15 is a flowchart showing a series of processes in the analysis apparatus according to the second modification. The analysis apparatus according to the second modification includes an intercepting unit, an extracting unit, a registering unit, a communication information holding unit, a classification unit, a classification result holding unit, and a counting unit. The intercepting unit intercepts the packet from the network (S502). The extraction unit extracts the start point address and the port number from the intercepted packet (S504). The registration unit registers the extracted start point address, port number, and packet data amount in the communication information holding unit (S506). The classification unit sorts the data held in the communication information holding unit with the start point address (S508). The classification unit classifies the sorted data into an original use and a non-original use (S510). The classification unit registers the classification result in the classification result holding unit (S512). If the predetermined condition is not satisfied (N in S514), the process returns to step S502. Thereby, the classification result is accumulated in the classification result holding unit until a predetermined condition is satisfied. When the predetermined condition is satisfied (Y in S514), the totaling unit totals the data accumulated in the classification result holding unit (S516).
According to the present modification, the amount of classification result data used as a total population can be adjusted by changing predetermined conditions. Therefore, it is possible to flexibly adjust the amount of classification result data according to the accuracy required for the index obtained as a result of aggregation.

  Although the present invention has been described based on the embodiments, the embodiments merely show the principle and application of the present invention, and the embodiments are defined in the claims. Needless to say, many modifications and arrangements can be made without departing from the spirit of the present invention.

  2 data center, 4 first subnet, 6 second subnet, 8 third subnet, 100 analysis device, 102 network interface unit, 104 communication information holding unit, 108 first index holding unit, 110 usage status acquisition unit, 112 usage status Holding unit, 114 coefficient holding unit, 116 first analysis unit, 118 second index holding unit, 120 display control unit, 122 interception unit, 124 extraction unit, 126 filter unit, 128 registration unit, 130 counting unit, 132 classification unit, 134 1st calculation part, 136 2nd calculation part, 138 Node information holding part, 140 Monitor, 142 1st synthetic | combination part, 144 Actual measurement characteristic acquisition part, 146 1st representative value calculation part, 148 Energy-saving degree score calculation part, 150 Actual measurement Characteristic holding unit, 152 Second analysis unit, 154 A second synthesis unit, 156, a second representative value calculation unit, and 158, a third index holding unit.

Claims (8)

An intercepting unit for intercepting communication information from the network including node information for identifying a node involved in communication and use information indicating the use of the communication;
An extraction unit that extracts node information and usage information from the communication information intercepted by the interception unit;
Communication information based on node information and usage information extracted by the extraction unit, communication information corresponding to the primary usage predetermined for the node, and communication information corresponding to non-primary usage that is a usage other than the primary usage A classification unit for classifying
An analysis unit that estimates the power used for the original use out of the total power used for the processing related to the communication based on the amount of communication information originally corresponding to the use. Analysis device. For a given node, a tabulation unit that tabulates the amount of communication information originally corresponding to the application,
A usage status acquisition unit that acquires information about the usage status of a predetermined node;
The analysis device according to claim 1, wherein the analysis unit associates the aggregation result in the aggregation unit with information on the usage status acquired by the usage status acquisition unit.   The filter unit according to claim 1, further comprising a filter unit that selects only communication information including predetermined node information based on the node information extracted by the extraction unit and outputs the selected communication information to the classification unit. The analysis device described in 1.   The analysis device according to claim 3, wherein the filter unit sequentially switches predetermined node information among node information of a plurality of nodes included in the network.   The characteristic acquisition unit for calculating an index related to the energy efficiency of the node by using an average value and a maximum value of parameters included in the usage status acquired by the usage status acquisition unit. The analysis device described in 1. Network,
An analysis device,
The analysis device includes:
An intercepting unit for intercepting communication information from the network including node information for identifying a node involved in communication and use information indicating a use of the communication;
An extraction unit that extracts node information and usage information from the communication information intercepted by the interception unit;
Communication information based on node information and usage information extracted by the extraction unit, communication information corresponding to the primary usage predetermined for the node, and communication information corresponding to non-primary usage that is a usage other than the primary usage A classification unit for classifying
An analysis unit that estimates the power used for the original use out of the total power consumption used for the processing related to the communication based on the amount of communication information originally corresponding to the use. Information processing system. Intercepting communication information from the network including node information for identifying a node involved in communication and usage information indicating the usage of the communication;
Extracting node information and usage information from the intercepted communication information;
Based on the extracted node information and use information, the communication information is classified into communication information corresponding to the original use predetermined for the node and communication information corresponding to the non-original use which is a use other than the original use. Steps,
An analysis that includes estimating the power used for the original use out of the total power used for the processing related to the communication based on the amount of communication information originally corresponding to the use. Method. A function of intercepting communication information from the network including node information for identifying a node involved in communication and usage information indicating the usage of the communication;
A function to extract node information and usage information from intercepted communication information;
Based on the extracted node information and use information, the communication information is classified into communication information corresponding to the original use predetermined for the node and communication information corresponding to the non-original use which is a use other than the original use. Function and
Based on the amount of communication information originally corresponding to the application, the computer is capable of estimating the power used for the original use out of the total power consumption used for communication processing. Computer program.

Images