JPS586183B2 - Data copy method between rotating storage devices - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention provides a method for copying data between rotary storage devices such as magnetic drums and magnetic disks, and particularly a rotary type that performs efficient data copying using a small capacity backup memory. This relates to a one-way data copy method between storage devices.

Rotary storage devices such as magnetic drums and magnetic disks are often used as external storage devices for computer systems.

Copying of information between drums and disks, which are the media of these storage devices, is performed for information protection and other purposes.

This is often done especially when the medium is a removable disk.

However, both are rotary storage devices, and their rotational phases generally do not match (synchronize).

For this purpose, various methods have been adopted in the past, one example of which is shown in FIG.

In Figure 1 a, 1 is a computer, 3 is an input/output interface 2
The disk control device connected to computer 1 via D
1 and D2 indicate magnetic disks, respectively.

The principle of this method is to use the normal exchange of information between a computer and a disk device as is, and perform copying through the intervention of a program.

Now assume that information is to be copied from D1 to D2.

First, track D stores the memory contents for one rotation of D1.
The information for one track is read by a program from the index indicating the beginning of
00).

This corresponds to t in FIG. 1b.

Next, the information for one track read above is written by the program to the track D2i corresponding to Dli of D2 from the beginning of the index of D2.

This corresponds to t2 in FIG. 1b. Similarly below, j3*j
4 is copied track by track.

In this method, one track is copied, so depending on the phase relationship of the index, a minimum of 2 rotations and a maximum of 3 rotations are required, with an average of 2.
It takes time for 5 rotations.

However, this method has the following drawbacks. That is, the intervention of a program is required for copying, which temporarily occupies the computer and input/output interface, and the operations of other programs and input/output devices are stopped during that time.

This impairs the online nature of computer systems because computers are partially involved despite online processing requests. should be avoided as much as possible if it involves online operations, which is a major drawback.

FIG. 2 shows five conventional examples that eliminate this drawback.

Reference numeral 3 indicates a disk control device, DI and D2 indicate magnetic disks, and 4 indicates a disk interface connecting the device 3 and disks D1 and D2.

The principle of this system is that the disk 1 is inside the disk controller 3.
Track buffer (portion 30 with hatching 4)
0), and the disk controller reads out one track's worth of information from D1 instead of the program (computer) shown in FIG. 1 described above, and temporarily stores it in the buffer.

This corresponds to 11 in FIG. 2b.

In general, it is not possible to overlap reading and writing between D1 and D2, and the contents of the buffer are written into D2 from the first index of D2 that comes after the storage in the buffer is completed.

This corresponds to 12 in FIG. 2b. One hour of copying using this method is the same as in the case of Fig. 1, with an average of 2.2.
It is 5 rotations.

The disadvantage of this method is that it requires a buffer for one track of the disk.

With recent advances in technology, the capacity for one track is 400.
The size of the buffer is increasing from 0 to 8000 bytes, and the burden on the hardware of this buffer is gradually increasing.

Particularly in devices that use the so-called sector method in which the recording unit is a fixed length, the price of this buffer accounts for 30 to 50% of the control device price, which is not economical, and this ratio is likely to increase in the future. It is in.

For this reason, a one-sided copy system with a buffer for one sector has been proposed, which will be explained below.

FIG. 3 is a diagram illustrating the operation timing of this system, especially the operation timing in the case of 12 sectors in one track.

First, the first sector of D1 is read (S1-R), and the read information is written after waiting until it becomes possible to write to the corresponding sector of D2 (SiW).

Next, read the second sector on D1 (S2-R), write to the corresponding sector on D2 (S2-W), and repeat the same procedure up to the last sector (12 sectors in the example of FIG. 3). Copying of the tracks is completed.

As is clear from the example of FIG. 3, in this method, the formal copying processing time is 123/12To, but the actual processing time is 13To, which requires the time equivalent to 13 rotations.

Generally, in the case of n sectors, time for n+1 rotations is required.

However, this time is unrelated to the rotational phase difference, and if the difference is small, this method is disadvantageous.

The present invention solves the above-mentioned drawbacks and provides a one-way data copying system between rotating storage devices that uses a small-capacity buffer memory to complete data copying in a short time. be.

The gist of the present invention is to provide a buffer memory for at least one sector, and rotate an intermediate work track with the buffer memory only when the rotational position is smaller than a predetermined rotational position difference between two storage devices. It is designed to virtually interpose data between the computer and the external device to perform data copying.

The optimal rotational position difference is 1/2 rotation, and in the following example, when the difference is smaller than 1/2 rotation, data copying is performed according to the above gist, and when it is larger than 1/2 rotation, data copying is performed by 1 rotation. I am trying to make a copy of the data below.

The present invention will be described in detail below.

FIG. 4 is a time chart explaining the operation timing of the present invention.

Assume that information on track i on disk 1 (Dli) is to be copied to track i on disk 2 (D2i),
D2j and D2k are work tracks on the disk 2.

Further, the number of sectors per track (12) and the rotational phase difference are the same as in the case of FIG.

First, the first sector on Dli is read (Sl-R), and after the reading is completed, writing is performed to the sector on D2j-rack that appears first (S1-W).

Next, the third sector on Dli is read (S3-R), and similarly written to the D2j track (83-W).

Thereafter, the 5th, 7th, . . . , 11th sectors are similarly copied in one rotation, and in the next rotation, the 2nd, 4th, .
. . , the 12th sector is copied in this order.

As a result, copying is performed on D2j with one sector shifted from D1i.

Therefore, next time, the information of D2j is copied onto DZk with a shift of one sector.

By repeating this, D2i and D2j or D2k
When the sector difference becomes 1, copying is performed to the corresponding sector on D2i, and the initial purpose of copying from Dli to D2i is completed.

The processing time until this copy is completed is 63/12To.

The time required for copying using this method varies depending on the rotational phase difference.

The minimum time is when copying from D1i to D2i is possible immediately, and it takes 21/2 revolutions.

On the other hand, the maximum time required is n transfers when the indexes of D1 and D2 are synchronized, which requires a time equivalent to 2n+1 rotations.

FIG. 5 shows a comparison of the copy processing time due to the rotational phase difference between the method of FIG. 3 and the method of the present invention (the method of FIG. 4).

As explained above, the system shown in FIG. 4 changes depending on the rotational phase difference as shown by the solid line l1.

On the other hand, the method shown in Figure 3 does not change depending on the rotational phase difference, and the dotted line l
As shown by 2, it always takes time for n+1 rotations.

That is, the present invention exhibits its effectiveness when the rotational phase difference is small.

Note that this rotational phase difference is "0" when D2 is shifted from D1 by one sector, and is "n" when the two disks are completely synchronized.

An embodiment of a circuit implementing the method of the present invention is shown in FIG.

The operation will be explained with reference to this figure and the time charts of FIGS. 3 and 4.

In FIG. 6, 1 and 2 are respectively disk 1 (DI),
Counters 3 and 4, which are counted up by the sector signal (pulse indicating the beginning of the sector) of the disk 2 (D2) and reset by the index, are used to count the number of written sectors. A detection circuit 6 is a subtraction circuit that performs subtraction between counters 1 and 2 and detects the rotation difference between D1 and D2. 7 is a down counter 8 that sets the rotation difference and decreases by 1 when the counter 5 finishes counting. is a detection circuit that detects that the difference output of the counter T is "1".

Reference numerals 9 indicate the output of the detection circuit 8, the output of a flip-flop 11 that stores whether the rotation difference is larger or smaller than 1/2 rotation, and the flip-flop 15 that is inverted every time the working tracks on D2 are moved between each other. As input, the three states of D
Address switching circuits that output select signals of I and D2 and track addresses of DI and D2, 10 to 15 are flip-flops, 16 to 19 are OR gates, 2
0 to 25 are AND gates, 26 is a buffer for at least one sector, and 27 is a suppression circuit that suppresses reading by one sector when 1/2 of one track is completed when writing between working tracks. respectively.

This figure will be explained in detail below, but it is assumed that the number of sectors per track is 12 and that information is copied from D1 to D2.

First, the case where D2 lags D1 by 1/2 revolution or more will be explained with reference to FIGS. 3 and 6.

By the start (copy start) signal shown in Figure 6, D1 and D
2 rotation difference (converted to the number of sectors) is counted as counter 1,
2 and its subtraction circuit 6, and sets it in the counter 7, and sets the flip-flop 11 when the difference is equal to or larger than 1/2 revolution.

On the other hand, the counter 5 that counts the number of copied sectors is reset through the gate 19.

Next, when the contents of counters 1 and 5 match by the matching circuit 3, since the flip-flop 10 has already been reset, the AND gate 22 and the OR gate are connected by sector 1 of D1.
The flip-flop 13 is set via the gate 11 to instruct D1 to read.

When sector 1 of D1 appears again, flip-flop 1
3 is reset, the read command is turned off, and the flip-flop 14 is set.

As a result, one sector worth of data of D1 is stored in the buffer 26.

This corresponds to the time for reading the first sector in FIG.

Next, when the coincidence circuit 4 detects that the counters 2 and 5 match, the flip-flop 12 is set by the sector 2 of D2, and a write finger (+) is output to D2.

When sector 2 of D2 appears again, flip-flop 1
2 is reset and writing of the data stored in the buffer 26 is completed.

This corresponds to the first sector write time in FIG.

At this time, the flip-flop 14 is reset and the counter 5 is counted up.

Further, when the contents of counters 1 and 5 match by matching circuit 3, flip-flop 13 is set by sector 1 for one sector, and data for one sector is stored in buffer 6 from D1.

This corresponds to the second sector read time in FIG.

Next, when the matching circuit 4 matches the contents of counters 2 and 5, sector 2 switches flip-flop 1 for one sector.
2 is set, and the data stored in the buffer is written to D2.

This corresponds to the second sector write time in FIG.

Thereafter, copying of data for one round of the track, that is, for 12 sectors, is executed in the same manner.

When the copying of data for one round is completed, the counter 5 overflows and outputs to the flip-flop 11 and A
Through the ND gate 24 and the OR gate 19, an initial state is entered, the track address is updated (not shown in the figure), and copying between the next tracks is started.

By doing the above for all tracks,
Copying from D1 to D2 is completed.

Next, when the rotation difference between D1 and D2 is smaller than 1/2 rotation,
Here, an example in which there is a rotation difference of three sectors will be explained below.

Similarly to the above, the start signal sets the rotation difference in the counter 1 and also resets the flip-flop 11 and the counter 5.

As a result, copying of the data of the track Dli on D1 to the work track D2j on D2 is started.

First, since the flip-flop 11 has been reset, the outputs of the coincidence circuits 3 and 4 are inhibited, and on the other hand, the Q output passes through the inhibit circuit 27, the OR gate 18, and the AND gate 23, and the flip-flop 13 is set to a settable state. becomes.

At this time, sector 1 is connected to AND gate 22 and OR gate 17.
The flip-flop 13 is set for one sector via .

As a result, one sector worth of data on Dli is transferred to the buffer 2.
6.

This corresponds to the first sector read time in FIG.

At this time, flip-flop 14 is set, so OR
Gate 16 and AND gate 20 enable flip-flop 12 to be set.

At this time, sector 2 causes flip-flop 12 to be 1
The data in the buffer 26 that has been set for sec is transferred to track D.
Written in 2j.

This corresponds to the first sector write time in FIG.

At this time, the flip-flop 14 is reset and at the same time the counter 5 is counted up.

This makes it possible to read the next sector on Di, and data for one sector is read out.

This corresponds to the read time of the third sector in FIG.

Copying is performed in the same manner thereafter, and when the copying for 1/2 rotation is completed, the reading is inhibited for a period of one sector by the inhibiting circuit 27 composed of a flip-flop and a gate circuit.

After this suppression period ends, copying is performed every other sector in the same way as above, and one rotation, or one track worth of data, is copied to the D.
is transferred from li to D2j and flip-flop 1
0 is set, 15 is also set.

In this state, the rotation difference between Dli and D2i that was originally present is D2j
This means that the size has been reduced by one sector between and D2i.

Therefore, the rotation difference counter 1 is decremented by 1 due to the overflow of the counter 5.

Next, copying from D2j to D2k is executed in the same manner.

In the example of FIG. 4, since the rotation difference is one sector at this time, the switching circuit 9 alternately selects tracks D2k and D2i by 8, and copying from D2k to D2i is started.

This copying is the same as the procedure described above except that the sector to which flip-flop 13 is set and reset is changed from sector 1 to sector 2 because flip-flop 10 is set.

This completes the copy from D1i to D2i, but
At this time, the AND gate 25 and the OR gate 19 return to the initial state, and the next track-to-track copy begins.

In the initial state, if the rotational difference between D1 and D2 is equivalent to a monthly sector, the detection circuit 8 and the switching circuit 9 act to directly copy from D1i to D2i.

The average time for one track copy according to the method of the present invention is as follows.

If the number of sectors per track is n, the maximum is n+1 rotation in the case of Figure 3, and the minimum is (2+-1/n) in the case of copying directly from Dli to D2i in Figure 4.
) rotation, and the average is 1/2+1/2+1/2n rotations, and since n is generally large, 12 to 32, it is approximately 1/n rotations.

Also, two working tracks, D2j and D2k, are required, but in the case of disks, such tracks are generally kept as spares, so this is not a big problem.
Even if you didn't use the analogy, the last two tracks can be compared to the fourth track.
The present invention can be applied according to the diagrammatic method.

According to the embodiment of the present invention, the buffer capacity is for one sector,
In other words, 1/n of one track is enough, and the total performance expressed as the product of one copy hour and buffer capacity is (n sectors) x (2.5 rotations)/(1 sector) x (1/
n rotations) -5.

Further, as in the conventional example shown in FIG. 1, this is the same as the conventional method shown in FIG. 2 in that no load is placed on the computer or input/output interface, and other conditions are also the same as in FIG.

Furthermore, since the rotation difference is checked every time one track is copied, it is possible to easily select the optimum method even if the rotation difference changes midway.

In the present invention, the 1/2 rotation difference is used as a standard, but this is theoretically optimal, and in practice, there may be slight variations depending on the circuit configuration.

Further, although the buffer memory is assumed to have one sector, it may have a capacity of one sector or more.

Furthermore, the copying method when the difference is larger than 1/2 rotation is not limited to the above embodiment.

[Brief explanation of drawings]

1, 8, b and 2 ayb are configuration diagrams and operation explanatory diagrams showing the conventional method, FIG. 3 is an explanatory diagram of the conventional method, FIG. 4 is an explanatory diagram of the method of the present invention, and FIG. FIG. 3 and FIG. 4 are graphs comparing the processing times of the systems, and FIG. 6 is a diagram showing an embodiment of the present invention. 1... Computer, DI, D2... Magnetic disk, 12... Flip-flop for write command, 13... Flip-flop for read command.

Claims (1)

[Scope of Claims] 1. First and second rotary storage devices employing a sector system in which the storage unit of its own information has a fixed length, and a buffer memory, wherein the first and second rotary storage devices are provided with In the data copying method between two rotary storage devices that copies data between two rotary storage devices, the first
, when the difference in rotational position between the second rotary storage devices is smaller than a predetermined rotational difference, the sectors corresponding to the capacity of the buffer memory are sequentially read from the first rotary storage device. Data is read out and temporarily stored in the buffer memory, and then read out from the buffer memory on one of up to two working tracks assigned in advance to a second rotary storage device or between the working tracks. The sector data is sequentially shifted and moved, and when the rotational difference with the corresponding track of the second rotary storage device reaches a sector corresponding to the capacity of the buffer memory, the sector data is sequentially moved to the corresponding track. A method for copying data between rotating storage devices. 2. First and second rotary storage devices employing a sector system in which the recording unit of its own information is a fixed length, and a buffer memory, wherein the first and second rotational storage devices are connected to each other via the buffer memory. In one type of data copying between rotating storage devices that copies data between
, when the difference in rotational position between the second rotary storage devices is greater than the difference of 1 month/2 rotations, data of sectors corresponding to the capacity of the buffer memory are sequentially read from the first rotary storage device and The data of the sector is temporarily stored in the buffer memory, and then the data of the sector read from the buffer memory is sequentially written and copied to the corresponding sector of the second rotary storage device, and the data is rotated between the first and second rotary storage devices. When the difference in position is smaller than the difference in months/2 rotations, data in sectors corresponding to the capacity of the buffer memory is sequentially read from the first rotary storage device and temporarily stored in the buffer memory, and then The sector data read from the buffer memory is sequentially shifted and moved between one area of a maximum of two working tracks assigned in advance to the rotating storage device or between the working tracks, and A disk copy unit between rotating storage devices in which data in a sector is sequentially moved and copied to a corresponding track when the rotational difference between the storage device and the corresponding track reaches a sector corresponding to the capacity of the buffer memory. method.