JPH0563817B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a control method for an external storage device such as a magnetic disk storage device in an electronic computer. A magnetic disk control device will be explained below as an example.

The connection configuration method for magnetic disk controller (DKC) and magnetic disk storage device (DKU) requires multiple units.
There is a device cross call configuration method in which multiple DKUs are connected to a DKC. Figure 1 shows the case where n DKUs are connected to two DKCs, 1 is the channel (CH) connected to the central processing unit, 2 is the channel (CH) connected to the central processing unit,
is DKC, 3 is the connection signal line, and 4 is DKU. By the way, in the conventional technology, during a series of control operations from the DKU positioning operation (SEEK operation), waiting for disk rotation, and the desired read/write (READ/WRITE) operation, the DKC commanded by the CH and The combination of DKUs was fixed, so even if you had multiple DKCs, you would not be able to fully utilize their capabilities. This will be explained using the case of FIG. 1 as an example.

Now, DKC-A is SEEK and
Assume that the SET SECTOR command is issued and the DKU enters a rotation wait state. In this state, DKC
-A is DKU-2 which gave the rotation wait command before that.
When the completion of rotation waiting (RECORD READY) is detected, data transfer (read/write operation) is started between the DKU-2 and the DKU-2. Therefore, after that DKU−
Even if 1 completes the rotation wait, DKC-A
is in a BUSY state with DKU-2, so
Unable to start data transfer with DKU-1. In other words, DKU-1 is receiving commands from DKC-A, and DKC-A is free (NOT
BUSY), data transfer cannot begin unless the status becomes RECORD READY. on the other hand,
DKC-B can only process the DKU for which it issued the SET SECTOR command. In this way, even if there are two DKCs, each DKC's ability cannot be fully demonstrated, and the DKU's RECORD READY
This will cause the signal to be missed.

An object of the present invention is to provide an external storage control method that eliminates the problems of the prior art described above and achieves higher performance.

In order to achieve the above object, the present invention eliminates the connection between an external storage control device and an external storage device, and connects the external storage device to any external storage control device by switching during a series of operation controls. By making it possible to transfer data using the external storage controller, the capabilities of the external storage control device/channel can be maximized.

An embodiment of the present invention will be described below with reference to FIGS. 2 to 8.

FIG. 2 shows an example of a connection configuration between DKC and DKU according to the present invention. In Figure 2, 2 is
DKC (3 units in this embodiment) and 4 are DKU (n units), and a connection switch 5 is installed between these so that any connection configuration can be established between the DKC and DKU. Note that the configuration shown in FIG. 1 may be used as long as DKC/DKU can be arbitrarily connected. each
The DKU has a control memory 6, which stores control information necessary for restarting control from any other DKC after the DKC/DKU connection is broken. FIG. 3 shows an example of the control memory 6, including channel numbers (CH NO),
DKC number (DKC NO), presence or absence of SET FN (command to set FN), SEEK ADDRESS,
SET SECTOR NO etc. are stored. In addition,
FM (FILE MASK) is information that determines the allowable range of seek and write operations for file protection.

Figure 4 shows an example of the configuration of the DKC, in which the CH interface control unit 10 performs interface control with the CH.
0, a device interface control unit 101 that performs interface control with DKC, a buffer memory 102 that temporarily stores transmitted and received data, a control memory 103 that stores microprograms necessary for control, and a CH interface control unit 100 that uses microprograms. , a device interface control 101, a microprocessor 104 that controls a buffer memory 102, and the like. Note that thin lines indicate control paths and thick lines indicate data paths.

Figures 5 to 8 show the control memory 10 in the DKC.
3 shows the processing flow of the microprogram stored in the microprogram 3, and the operation will be explained below using these.

Figure 5 shows that DKC is activated from the channel,
It is a flowchart of control from merging specified DKUs to separating them. When the DKC 2 receives input/output commands from the channel through the CH interface control unit 100, the microprocessor 10
4, connects the specified DKU 4 through the device interface control unit 101, and
Controls commands and responses between CH/DKU (steps 10 and 11). The operation when DKC/DKU is in a combined state is the same as before. During a series of operation controls, it is tested whether the DKU4 can be disconnected (step 12), and if it is possible to disconnect it (for example, SEEK and SET are applied to the DKU4).
When the SECTOR command is issued and the DKU enters the rotation wait state, the DKC 2 transmits the control information (contents in FIG. 3) stored in the control memory 103 of the DKC through the device interface control unit 101. After transferring the data to the control memory 6 of the DKU 4, the DKU 4 is logically separated (steps 13 and 14). As a result, the DKC2 becomes free and waits for the next activation from the CH or a connection request from any DKU.

FIG. 6 is a flowchart of control after the DKU becomes necessary to connect with the DKC due to completion of SEEK, completion of rotation waiting, etc., for example. When a join request is made from DKU to DKC, the join request is
It is sent to all DKCs that have a connection path to the DKU, but among them, the DKC that is in a free state enters the selection operation for the DKU in question (step 20). If multiple DKCs perform a selection operation, the one that started the selection operation first
The DKC completes the connection with the DKU, and the DKC that fails in the connection returns to the free state (step 21).
Whether or not the DKC/DKU combination has been established is determined based on, for example, a response from the DKU to the connection try operation. Once the connection with DKU is established, DKC2
is under the control of the microprocessor 104.
The control information stored in the control memory 6 of the DKU 4 is imported into the control memory 103 through the device interface control unit 101 (step 22);
After returning to the state before DKU separation, a request to join the channel will be issued. On the other hand, as a channel, it is possible to use either or both of a conventional channel without a disconnection function (referred to as a conventional channel), a channel with a disconnection function similar to this DKC (referred to as a new channel). Depending on the capability of the channel, in the case of a conventional channel, a connection request is issued through the CH interface control unit 100 only for the channel that issued the startup request to the DKC (step 23), and once the connection is established, the control is started. Perform restart (step 24). Also,
In the case of a new channel, a connection request is issued to all channels through the CH interface control unit 100 (step 25), and the subchannel information of the channel that was previously combined is sent to the channel that has been combined. channel no., device
Transfer NO (steps 26, 27) and allow operation to continue. In order to be able to switch operations in this channel, it is necessary to share subchannel registers between a group of channels, but this is fully achievable with advances in LSI technology.
The processing of the DKC2 after the control is restarted is the same as that shown in FIG.

FIGS. 7 and 8 show still another embodiment for changing channels. In this embodiment, when a CH/DKU is disconnected, the DKU requests the channel concerned to transfer the contents of the subchannel register (step 32), and not only the contents of the DKC control memory necessary for the DKC to resume control, but also The contents of the subchannel register required for the channel to resume control are kept in standby in the corresponding DKU (step
33) When recombining with a channel, in addition to the previously connected channel number and device number, the contents of the subchannel register are also transferred to the channel to which the connection was established (step 42). Resumes operation. In this example,
Since all the information necessary to resume operation in the combined channel is retrieved via the device, there is no need to share subchannels between channels.

Next, a specific explanation will be given of how much the recombination failure rate with channels during RECORD READY is reduced when the present invention is applied, compared to the prior art.
Now, channels and DKC are tied together (one-to-one correspondence), DKU is accessed equally on each route, and the number of channels (= number of DKC) N _c = 4, number of DKU
N _d 16, channel (=DKC) usage rate ρ _CH = 0.3, recombination failure probability ρ _b at RECORD READY
In the conventional technology, ρ _b =N _d -1/N _d -ρ _CH 0.3, but when the present invention is applied, ρ _b = (N _d -1/N _d -ρ _CH ρ _CH ) ^Nc (0.3) ⁴ 0.01. In other words, the channel recombination failure probability is reduced to 1/30, which becomes an almost negligible value, and a significant improvement is possible.

Furthermore, the effects of the present invention will be even more significant if the DKC and channel are not connected one-to-one. In other words, in general, even if DKC becomes free, it does not necessarily mean that the channel is free, and in the case of traditional DKC and channel connections, the combination would not be established unless both the channel and DKC, which have a one-to-one correspondence, are free. However, in the embodiment, if there is at least one free device in each of the DKC and the channel, the connection will be established.

Furthermore, in the present invention, even if the DKC or channel stops functioning due to a failure during operation, the remaining
Continued operation is possible on DKC/channel,
Significant improvements in functionality and reliability can be expected in terms of troubleshooting.

The present invention has been explained above using DKC as an example, but
Of course, it goes without saying that the present invention is not limited to this.

[Brief explanation of drawings]

FIG. 1 is a diagram showing an example of the connection configuration between a conventional DKC and DKU, FIG. 2 is a diagram showing an example of the connection configuration according to the present invention, and FIG. 3 is a diagram showing an example of the configuration of a control memory provided in the DKU. FIG. 4 is a diagram showing an example of the hardware configuration of the DKC, and FIGS. 5 to 8 are flowcharts for explaining the control operation of the present invention. 1... Channel, 2... Magnetic disk controller (DKC), 3... Connection signal line, 4... Magnetic disk storage device (DKU), 5... Connection switch, 6...
DKU control memory.

Claims (1)

[Scope of Claims] 1 A system comprising a plurality of external storage control devices and a plurality of external storage devices mutually controlled by the plurality of external storage control devices, the external storage control device being connected to a central processing unit through a channel. In the computer system, a control memory is provided corresponding to the external storage device, and each of the external storage control devices resumes control at a break in a series of operation controls for the external storage device specified by the channel. A function that transfers the control information necessary for the external storage device to the control memory corresponding to the external storage device and disconnects the external storage device, and a function that disconnects the external storage device if it is not in operation when a connection request is generated from any external storage device. 1. An external storage control method comprising a function of selecting an external storage device that has received a connection request, importing control information from a control memory corresponding to the external storage device, and restarting control. 2. When resuming control, the external storage control device issues a connection request to a plurality of channels, selects any channel that is not in operation at that time, and resumes control. External storage control method.