JP3094941B2 - OSI Network Management Protocol Performance Analysis System - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an OSI network management protocol performance analysis system, and more particularly to an OSI network management protocol performance analysis system capable of analyzing the performance of a complex and large-scale system based on queuing theory.

[0002]

2. Description of the Related Art Conventionally, in this type of OSI network management protocol performance analysis method, there is an urgent need for performance estimation in advance in order to increase the demand for OSI network management due to system complexity and large scale and to efficiently operate the OSI network. It is used because it has become. An example of a calculation method in the conventional OSI network management protocol performance analysis method is described in Information Processing Society of Japan, 1983, Vol. 24
No. 33 pp. 326-334, "Methods for calculating a mixed queuing network". The calculation method described in this academic journal is a calculation method in performance analysis when there are a plurality of closed chains and open chains at the same time.

[0003] Embodiments of the conventional example will be described in detail with reference to the drawings.

FIG. 10 is a flowchart of a conventional example.

Referring to FIG. 10, in a conventional embodiment, a modeling unit 1 for modeling a target OSI network management protocol in a queuing network, and a "combined queuing network calculation method" is described in the modeling result. A performance analysis unit 31 that applies and performs a performance analysis, an average guest number calculation unit 41 that calculates the average number of guests of each node from the performance analysis result, and a traversing time of each chain is calculated from the result of the performance analysis unit 31. And a traversing time calculation unit 51.

The modeling unit 1 classifies the OSI network management protocol into a data protocol and a fixed-cycle protocol based on its characteristics, and models the data protocol into a closed chain and the fixed-cycle protocol into an open chain.
The data protocol indicates a transport layer data protocol defined by OSI and its confirmation data, and the fixed cycle protocol indicates a protocol periodically generated in the network layer defined by OSI. In the queuing network model, the data protocol path is a closed chain, and the number of customers is modeled by the number of windows specified in the system. The path of the fixed-cycle protocol is an open chain, and the arrival rate to the chain is modeled by the reciprocal of the generation cycle of the fixed-cycle protocol. In addition, a node such as a package on a route is set as a service center, and the processing time required by the node when there is no influence from other protocols is modeled in correspondence with the average service time. At this time, the load amount is obtained by multiplying the average service time for the service center by the average number of visits.

[0007] The performance analysis unit 31 applies the “calculation method of the mixed queuing network” to the model modeled by the modeling unit 1 to obtain the average number of guests and the average stay time in each service center in each chain.

[0008] The average number of staying guests calculation unit 41 is
The average number of guests staying at each service center is calculated from the analysis result of (1).

The traversing time calculator 51 calculates a traversing time, which is a time required for each chain to traverse the route, from the analysis result of the performance analyzer 31.

Next, the operation of the conventional embodiment will be described with reference to FIG.

The modeling unit 1 models a data protocol in an OSI network management protocol by a closed chain and a fixed cycle protocol by an open chain. First, a method of modeling closed chains will be described. The following symbols are used in this description.

WI indicating the number of windows defined in the system, service center n by data protocol d
Tdn, representing the average service time of the data protocol d
Represents the average number of visits to the service center n, and DWdn represents the load of the data protocol d on the service center n.

[0013] In the data protocol which is a closed chain, the number of customers P of each data protocol is P since the number of customers of the system predetermined window number WI is always circulating on the route.
= WI modeled. The load amount DWdn to each service center n is modeled as DWdn = Tdn × θdn (1).

Next, a method of modeling an open chain will be described. The following symbols are used in this description.

Gt representing the generation cycle of the fixed cycle protocol t which is an open chain, λ representing the arrival rate of the fixed cycle protocol
t = 1 / Gt, Ttn representing the average service time of the service center n by the fixed cycle protocol t, ηt representing the average number of visits to the service center n by the fixed cycle protocol t
n, TWtn indicating the load of the fixed-cycle protocol t on the service center n, and the above five types of symbols are used.

The arrival rate λt of the fixed cycle protocol t which is an open chain is modeled by λt = 1 / Gt. The load amount TWtn to each service center n is modeled by TWtn = Ttn × ηtn (2).

The performance analysis unit 31 performs a performance analysis on the result of the modeling unit 1 by applying the “computation method of the mixed queuing network”. The result sought here is that each data protocol d, which is a closed chain, has an average number of staying guests DPdn representing the total number of how many customers are waiting for processing at each service center n and how many customers are processing. Similarly, the average stay time DSdn, which is the average of the time during which each data protocol d is waiting for processing at the service center n and the time during which processing is being performed, and each fixed cycle protocol t, which is an open chain, is also The average number of staying customers TPtn representing the total number of customers waiting for processing at the service center n and the number of customers being processed, and the time each fixed-cycle protocol t waits for processing at the service center n And the average stay time TStn, which is the average of the sum of the times during which the processing is being performed.

The average number of guests calculating unit 41 calculates each service center n from the results DPdn and TPtn of the performance analyzing unit 31.
Is obtained as follows.

Pn = ΣDPdn + ΣTPtn (3) (where, in the Σ, the first term represents the sum for all data protocols d, and the second term represents the sum for all fixed-cycle protocols t.) , The result DSd of the performance analysis unit 31
traversal time CT of each data protocol d from n and TStn
d and the traversing time CTt of each fixed cycle protocol t are obtained as follows.

CTd = ΣDSdn (4) (where Σ represents the sum for all service centers n) CTt = ΣTStn (5) (where Σ represents the sum for all service centers n) The performance analysis results of the OSI network management protocol in the conventional example are obtained as the average number of guests Pn (Equation (3)) at each service center n and the traversing time CTd (Equation (4)) of each data protocol d. , Traversing time CTt (Equation (5)) for each fixed-cycle protocol t.

[0021]

The above-mentioned prior art has the following problems.

The first problem is that performance analysis of a complicated and large-scale system cannot be performed.

The reason is that such calculation is usually performed by a program on a computer, but when the number of paths, that is, the number of protocols increases, the time and memory required for calculation increase exponentially, and the calculation becomes complicated. This is because a large-scale system may not be usable.

The second problem is that when performing model conversion to reduce the number of routes by combining routes in order to reduce the time and memory required for the calculation to a range that can be calculated on a computer, the In other words, the traversing time for each route before combining is not determined.

The reason is that, in order to apply the “computation method of the mixed type queuing network” after the model conversion, the average residence time obtained from this is for the combined route after the conversion and corresponds to the route before the conversion. This is because it is not for each protocol. Therefore, the traversing time represented by the sum of the average stay time of each protocol with respect to all nodes is not obtained for each protocol.

In view of the above-mentioned problems of the prior art, an object of the present invention is to provide an OSI network management protocol performance analysis system for finding the average number of guests in each service center based on queuing theory in a complex and large-scale system. To provide.

Further, another object of the present invention is to determine the traversal time of each protocol with respect to the route before conversion in addition to the average number of guests staying at each service center so as to obtain the same result as the conventional example. An object of the present invention is to provide a performance analysis system that is required even if a model conversion is added.

[0028]

SUMMARY OF THE INVENTION An OSI network management protocol performance analysis system according to the present invention has
A modeling unit for modeling the SI network management protocol in a queuing network, and a data protocol group, which is a closed chain of the OSI network management protocol, modeled by the modeling unit, are obtained by calculating the number of customers and the load on each service center. Into two closed chains, and
The fixed-period protocol group, which is an open chain of OSI network management protocols modeled by the modeling unit, is converted into one open chain by calculating the arrival rate and the load on each service center, and performance analysis is performed on a computer. A "mixed queuing network calculation method" is applied to each of the closed and open chains converted by the model conversion unit, and closed at each service center. Average number of visitors in the type chain and average number of visitors in the open type chain,
And a performance analysis unit for calculating the average residence time of the closed type chain and the average residence time of the open type chain;
From the sum of the average number of guests in the closed chain and the average number of visitors in the open chain at each service center, the average number of guests at each service center in the entire system is obtained. Average visitor number calculation unit for calculating the analysis result.

That is, the model conversion unit converts the data modeled by the modeling unit into only one closed-type chain and one open-type chain. Based on the conversion result, the performance analysis unit applies the “combined queuing network calculation method”, but since there are only two chains, the time and memory required for the calculation can be reduced.

Also, in the above-mentioned OSI network management protocol performance analysis system, based on the average staying time and the load amount of the closed chain in each service center, processing is performed in each service center in the closed chain after conversion. The average waiting time required for each data protocol is calculated from the average waiting time and the load on each service center of each data protocol, and the average staying time and load of the open chain at each service center are calculated. From the amount, the average waiting time required to receive processing at each service center in the open type chain after conversion is obtained, and from the average waiting time and the load on each service center of each fixed cycle protocol, each fixed cycle protocol is calculated. A traversing time calculation unit is provided to calculate the traversing time of the It may be.

That is, the traversing time calculation unit can calculate the traversing time of each protocol before conversion from the analysis result of the performance analysis unit.

[0032]

Next, embodiments of the present invention will be described in detail with reference to the drawings.

(First Embodiment of the Present Invention) FIG. 1 is a flow chart of the first embodiment of the present invention.

Referring to FIG. 1, the configuration of the present invention is a modeling unit 1 for modeling a target OSI network management protocol in a queuing network, and converts the modeling result into one closed chain and one open chain. A model conversion unit 2 and a performance analysis unit 32 for performing a performance analysis by applying a “combined queuing network calculation method” to the model conversion result
And an average number of guests calculating unit 4 for calculating an average number of guests of each node, that is, each service center from the performance analysis result.
2 and the operation is in the order of the flow of FIG.

The modeling unit 1 is the same as the conventional modeling unit 1.

The model conversion unit 2 converts a data protocol group, which is a closed chain of the OSI network management protocol modeled by the modeling unit 1, and a fixed-period protocol group, which is an open chain, into one closed chain and one open chain, respectively. Convert to

The performance analysis unit 32 applies the “computation method of the mixed type queuing network” to the conversion result of the model conversion unit 2 and
The average number of guests and the average length of stay at each service center in the closed chain and the open chain are determined.

The average number of staying guests calculation unit 42 is
The average number of guests staying at each service center is calculated from the analysis result of Step 2.

Next, the operation of the embodiment of the present invention will be described with reference to FIG.

The modeling unit 1 models the data protocol in the OSI network management protocol with a closed chain and the fixed-period protocol with an open chain. The model conversion unit 2 converts each of the two types of closed type chain and open type chain existing as a result of the modeling unit 1 into one. First, a method of converting a closed chain will be described. The following symbols are used in this description.

The above three types of symbols: WI indicating the number of windows defined in the system, DN indicating the number of paths of the data protocol which is a closed chain, and DWdn indicating the load on the service center n by the data protocol d. Is used.

The number of customers P in the closed chain after the conversion is obtained as follows.

P = WI × DN (6) The load DTn to the service center n after the conversion is obtained as follows.

DTn = (ΣDWdn) / DN (7) (where Σ represents the sum of all data protocols d.) Conversion to one closed chain is performed by the number of customers P
(Equation (6)) and the load amount DTn to each service center n
(Equation (7)).

Next, a method of converting an open chain will be described. The following symbols are used in this description.

Gt indicating the generation cycle of the fixed cycle protocol t which is an open chain, λt = 1 / Gt indicating the arrival rate of the fixed cycle protocol t, TWtn indicating the load of the fixed cycle protocol t on the service center n, The above three types of symbols are used.

The arrival rate λ of the converted open chain is obtained as follows.

Λ = Σλt (8) (where Σ represents the sum of all fixed-cycle protocols t.) The converted load TTn to the service center n is obtained as follows.

TTn = Σ (λt / λ × TWtn) (9) (where Σ represents the sum of all fixed-cycle protocols t.) Conversion to one open chain is performed by arrival rate λ (( 8) and the amount of load TTn to each service center n (equation (9)).

In the performance analysis unit 32, the number of customers P, which is the result of the model conversion unit 2, and the load amount DTn to each service center n
, And one open chain after the conversion represented by the arrival rate λ and the load TTn to each service center n. Perform a performance analysis by applying the "method". The result obtained here is the average number of guests DP in the closed chain service center n.
n and the average stay time DSn, the average number of guests TPn and the average stay time TSn of the open-type service center n.

The average number of staying guests Pn is calculated from the result of the performance analysis unit 32 by the average number of guests Pn of each service center n.
Is determined as follows.

Pn = DPn + TPn (10) In the first embodiment, the performance analysis result of the OSI network management protocol is obtained from each service center n.
Is the average number of guests Pn (equation (10)).

Next, the operation of one example of the first embodiment of the present invention will be described in detail.

FIG. 2 is a diagram showing an embodiment of the present invention.

Referring to FIG. 2, in this model, there are four types of data protocols a and b, which are closed chains, and fixed cycle protocols c and d, which are open chains. The route of each protocol is indicated by each arrow.

This embodiment is modeled by the modeling unit 1 as shown in FIG. In FIG. 3, the numbers surrounded by circles represent service center numbers, and the symbols above the circles represent the average service time at the service center. The arrows correspond to FIG. 2, and the flow of the arrows indicates the path of each protocol. The table also summarizes the load of each protocol at each service center. The bottom line shows the number of customers in the data protocol and the arrival rate in the fixed-cycle protocol.

When FIG. 3 is model-converted by the model converter 2, the result of converting the data protocols a and b into one closed chain is shown in FIG. 4, and the result of converting the fixed-period protocols c and d into one open chain. Is as shown in FIG. The conversion to the closed type chain is performed by the load amount D to each service center according to the equation (7).
Tn is obtained, and the number of customers is obtained by using equation (6). For conversion to an open chain, the load TTn to each service center is obtained from equation (9), and the arrival rate λ is expressed by (8)
It can be obtained using an equation.

FIG. 6 shows the performance analysis of FIGS. 4 and 5 by the performance analysis unit 32. The “mixed queuing network calculation method” is applied to each of the closed chain and the open chain converted by the model conversion unit 2 to obtain an average number of guests DPn of the closed chain in the service center n and Average number of guests TPn in open chains and average staying time D in closed chains
The average stay time TSn of Sn and the open type chain is obtained.

FIG. 7 is calculated by the average number of guests calculating unit 42 as shown in FIG. From the result of the performance analysis unit 32, the average number of guests Pn of each service center n in the entire system is obtained from the sum DPn + TPn of the average number of guests in the closed chain and the open chain of each service center n.

The result of the performance analysis according to the first embodiment of the embodiment is the average number of guests Pn of each service center n.

(Second Embodiment of the Present Invention) Next, a second embodiment of the present invention will be described in detail with reference to the drawings.

FIG. 8 is a flowchart of the second embodiment of the present invention.

Referring to FIG. 8, in the second embodiment of the present invention, in addition to the configuration from modeling unit 1 to average number of guests calculating unit 42 in the first embodiment shown in FIG. The difference is that a time calculation unit 52 is provided.

The traversing time calculation section 52 calculates the traversing time of each data protocol and each fixed-cycle protocol from the analysis result of the performance analysis section 32.

Next, the operation of the second embodiment of the present invention will be described in detail with reference to FIG.

The modeling unit 1, the model conversion unit 2, the performance analysis unit 32, and the average number of guests calculating unit 42 are respectively the modeling unit 1, the model conversion unit 2, the performance analysis unit 32, and the average number of guests in the first embodiment. The description is omitted because it is the same as the calculation unit 42.

In the traversing time calculating section 52, the performance analyzing section 32
The traversing time required for each data protocol and each fixed-period protocol to traverse the route is obtained from the result of the following. First, in the closed chain after conversion, the service center n
Is expressed as DWn, the average waiting time DSn and the load amount DTn in the converted closed service center n are obtained as follows.

DWn = DSn-DTn (11) When the average waiting time at the service center n in the open chain after conversion is represented by TWn, the average stay time TSn and the load at the service center n in the open chain after conversion are represented by TWn. From the quantity TTn, it is obtained as follows.

TWn = TSn-TTn (12) From equation (11) and the load DWdn of the data protocol d on the service center n, the traversing time CTd of the data protocol d is obtained as follows.

CTd = Σ (DWn + DWdn) (13) (where Σ represents the sum of the service centers through which the data protocol d passes among all the service centers n. Therefore, when DWdn = 0, DWn + DWdn = 0.
And ) Also, (12) and the service center n of the fixed cycle protocol t
From the load amount TWtn, the traversing time CTt of the fixed cycle protocol t is obtained as follows.

CTt = Σ (TWn + TWtn) (14) (where Σ represents the sum of service centers passing through the fixed-period protocol t among all service centers n. Therefore, when TWtn = 0, TWn + TWtn = 0.
And The traversal time calculation unit 52 determines the traversal time CTd of each data protocol d as the performance analysis result of the OSI network management protocol of the second embodiment ((13)
Equation) and the traversing time CTt ((1
4) Expression).

Next, an embodiment will be described. When the performance analysis of the example shown in FIG. 2 is performed in the second embodiment shown in FIG. 8, the results from the modeling unit 1 to the average number of guests calculating unit 42 are the same as those in the first embodiment. Is the same. Therefore, when the calculation is performed by the traversing time calculation unit 52 using the result of FIG. From the result of the performance analysis unit 32, the traversing time of each data protocol and the traversing time of each fixed cycle protocol are obtained.

The results of the performance analysis according to the second embodiment of the present invention are the average number of guests Pn of each service center n, the traversing time CTd of each data protocol, and the traversing time CTt of each fixed-cycle protocol.

[0074]

As described above, the present invention firstly provides:
Performing model conversion to reduce the number of paths has the effect of enabling performance analysis by applying the “computation method for a mixed queuing network” to a complex and large-scale system.

The reason is that the time and memory required for the calculation, which are restrictions when the calculation is actually performed by the computer, are kept within an executable range.

The second effect is that the traversal time of each route before the model conversion, that is, the traversal time of each protocol, can be obtained from the result of the performance analysis to the result of the model conversion for reducing the number of routes.

The reason is that, based on the average staying time per service center for the data protocol group and the fixed-period protocol group which are respectively grouped by the model conversion, the data protocol before the conversion and the service center for each fixed-period protocol are converted. This is because it has become possible to calculate the average stay time and calculate the crossing time from the sum of them.

[Brief description of the drawings]

FIG. 1 is a flowchart of a first embodiment of the present invention.

FIG. 2 is a diagram showing an embodiment of the present invention.

FIG. 3 is a diagram illustrating a result of a modeling unit according to the embodiment.

FIG. 4 is a diagram illustrating a result of a closed chain among results of the model conversion unit according to the embodiment.

FIG. 5 is a diagram illustrating a result of an open chain among results of a model conversion unit according to the embodiment.

FIG. 6 is a diagram illustrating a result of a performance analysis unit according to the embodiment.

FIG. 7 is a diagram illustrating a result of an average staying guest number calculation unit according to the embodiment.

FIG. 8 is a flowchart of a second embodiment of the present invention.

FIG. 9 is a diagram illustrating a result of a traversing time calculation unit according to the embodiment.

FIG. 10 is a flowchart of a conventional example.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Modeling 2 Model conversion part 31, 32 Performance analysis part 41, 42 Average visitor number calculation part 51, 52 Crossing time calculation part

──────────────────────────────────────────────────続 き Continuation of front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H04L 12/26 H04L 29/06 H04L 29/14 G06F 13/00 351

Claims (2)

(57) [Claims] 1. A modeling unit for modeling a target OSI network management protocol in a queuing network, and a data protocol group, which is a closed chain of the OSI network management protocol, modeled by the modeling unit, the number of customers and The amount of load on the service center is calculated and converted into one closed chain. The fixed-cycle protocol group, which is an open chain of the OSI network management protocol, modeled by the modeling unit is used as an arrival rate and each service. A model converter for calculating the load on the center and converting it into one open-type chain to enable performance analysis on a computer; and a closed-type chain and an opener for each one converted by the model converter. Apply the “combined queuing network calculation method” to the type chain, and calculate the average A performance analysis unit for calculating the number of customers and the average number of guests staying in an open chain, and the average stay time of a closed chain and the average stay time of an open chain, and a closed type in each of the service centers obtained by the performance analysis unit. From the sum of the average number of guests staying in the chain and the average number of guests staying in the open chain, the average number of guests of each service center in the entire system is obtained, and the performance analysis result of the system is calculated from the analysis result for the converted modeling. An OSI network management protocol performance analysis system, comprising: 2. The OSI network management protocol performance analysis system according to claim 1, wherein each of the services in the closed chain after the conversion is converted from an average stay time and a load amount of the closed chain in each service center. Calculating an average waiting time required to receive processing at the center; determining a traversing time of each data protocol from the average waiting time and a load amount of each data protocol on each service center; From the average stay time and the load amount of the open chain in the above, the average waiting time required to receive the processing in each of the service centers in the converted open chain is calculated, and the average waiting time and each of the fixed cycle protocols are obtained. From the load on the service center, the traversing time of each of the fixed-cycle protocols is obtained, and the OSI network management protocol performance analysis system comprising the cross-time calculation section for calculating a performance analysis results.