TWI296776B - - Google Patents

Description

1296776 · IX. DESCRIPTION OF THE INVENTION: TECHNICAL FIELD OF THE INVENTION The present invention relates to a logical disk management method, and relates to a logical disk snapshot backup management method. [Prior Art] Logical Volume Manager (LVM) is a method of dividing the hard disk space into "elastic" logical disks. With LVM technology, the hard disk can be simply re-partitioned. Re-sizing. LVM gives us the flexibility of disk management. The flexibility of TM is that it can be composed of several different types and sizes of physical disks and partitions into multiple logical disks (Logical V〇l_, LV), that is, virtual partition management. With __ __ i TM provides a higher level of disk storage for the computer, multiple partitions of the hard disk should be, and one is managed by the disk group. The storage disk can be used to improve the partition size and removal as needed. It can be used to make full use of the space. It allows the secret administrator to more easily allocate storage for Ying Jun: Manage storage disks by user group, allowing administrators to identify storage disks with more straightforward names instead of physical disk names. LVM - A very nice feature is snapshots ((4) coffee (8). Snapshot technology is a kind of logic Disk Backup mode, you can. 攸, album disk creation, the purpose is to retain the original disk at a certain point in time data content. It ^ » Start disk to save the form of the storage room _ _ reduced design, it scales Through the complete cloning of the f material into the 1296776 photo: record the original value of the change position in the corresponding logical disk, so the advantage of achieving inter-speed, low space consumption -... The data that changes with time retains different time points:: Technology; can be based on the requirements of the forest _ _ _ build (four), Miao fast Zhao through the leaves to present the image of the disk in the fast-off scale, Ke _ version, regardless of the state of the original disk has now become. Snapshot What the backup does is not a complete cloning of the data, but a technique called “write-on-write” (〇w On Write, C0W) to implement the backup. For the original ones with one or more snapshots, when there is: “Please tree First, the original content on the address to be written is backed up to the snapshot empty, and then the original logic is performed. When the initial (10) operation is completed, and the original address is written, the original address is written. No longer execute in this snapshot C〇w operation. In this case, the snapshot retains the data on some addresses of the original disk when it was created. When accessing a snapshot, you need to get the version of the original disk at the corresponding time point of the snapshot. Using # snapshot technology to read the unchanged information from the original disk at the corresponding time point, and reading the original data under the changed address from the snapshot, and then integrating the original of the corresponding time point The full version of the disk. _ In the use of the logical disk, not all the data of the address will be modified. Therefore, the quick makeup space can be smaller than the corresponding logical disk space, which can meet the usual requirements. The advantage of low idle consumption for clone backups, but also, if the number of modified addresses on the logical disk exceeds the snapshot space, the snapshot will not finish writing 1296776

The main risk comes from the risk of space. Causes the snapshot to fail. Therefore, the snapshot is invalid. Since the original disk may be modified or written, it needs to be in multiple

Different days, the custodial point is fast (4) * There will be a lot of repeated phenomena, so the usage of each snapshot is increased, which also increases the space occupancy of the overall snapshot, increasing the loss, the knife, the 逑 and the ' This affects the efficiency of logical disk data backup; the other is because the logical disk is independent of each other, and the effect is effective. SUMMARY OF THE INVENTION In view of the above problems, the main object of the present invention is to provide a snapshot management method for a logical disk manager, which not only functions as a high-speed backup of the county, but also saves the storage space capacity of the snapshot. Therefore, in order to achieve the above objective, the logical disk management crying snapshot management method disclosed in the present invention includes the following steps: performing a pre-write copy on the logical disk in the case where a plurality of snapshots are corresponding to a logical disk. During operation, the data address of the data disk to be written in the logical disk is searched; in the snapshot established at different time points, the snapshot created at the last time point is written according to the pre-write copy to the upcoming data 1296776. The data under the data address is backed up; when there is a snapshot read request, it is judged whether there is a request data address in the snapshot corresponding to the read request; and when the snapshot does not exist in the snapshot, the request data address 'and the snapshot location If the corresponding time point is not the last time point, then look for a plurality of snapshots later than the corresponding time point of the snapshot. ^Find out whether there is a request data button for each fast money, until the search for the data address is found. Snapshot data under the data address. According to the snapshot management method of the logical disk manager of the present invention, when there is a request data address in the fast display, the snapshot under the request data address is directly read ^^ a snapshot of the logical disk manager according to the present invention The management method, wherein t does not "request the data address, and the time point corresponding to the snapshot is the last %", then H takes the original data under the logical disk capping address. According to the logic snapshot management method of the present invention, Remove a quick snapshot from a plurality of snapshots. Rjnli A. Eight to be - the earliest time / exposure of the ride should be taken at a time _ section, the job deletes the fast money corresponding to the miscellaneous π in its previous snapshot. Add to the pair, Luo single magnetic axis (four) has _establish __, t perform a pre-write operation on the logical thin disk, ^ system, in the process of 'so not only save the snapshot need to save the storage ^ write The effect of the continuation of his shirt is low. (4) In addition to the technique of 'Axis (4), the through-side merging algorithm will require 1296776 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Spear, 喿作Q this 'will not be due to the logic of the invention The snapshot of the disk manager's snapshot tube 2 method, so that the deletion of a certain snapshot causes the other snapshots to be wrong. [Embodiment] The embodiment details about the present invention _ features and implementation, !! The drawings are best described below. Before the description of the snapshot management method of the logical disk manager of the present invention, the conventional snapshot backup method will be described with reference to the drawings. "Pic 1" is the second data to be written at the logical disk at different points in time: =^ schematic diagram; "^^" is a schematic diagram of the snapshot according to the prior art at different time points; 2b diagram" is the diagram; and "ΐΓ: the second snapshot of the snapshot backup method is at a different time point; the brother map" is the schematic diagram of the third snapshot of the snapshot backup method according to the prior art. The figure is the original memory of the logical disk corresponding to the snapshot at different time points. Figure 1 is a schematic diagram of the data to be written at the different time points and the contents of the poor material. As shown in the figure, the logic block The area 10, 2 〇 and the knife J are the logical data address and its corresponding original data content between the time points tl, t2 and t3. +1296776 is written. It is assumed that the addresses of the logical disk at time t and their corresponding original data contents are as shown in the data area 301 of "Fig. 3", and "Fig. 3" corresponds to the snapshots at different time points. Schematic diagram of the original data of the logical disk, which is given in the form of a table 3〇 composed of two-dimensional variables of time t and logical address. The current Journey (four) materials are recorded and the scales are; No information was recorded in the book. If there is a data change at a certain time point tr after the snapshot is established, that is, tl ' <U time point logical disk, for example, the data under the data address is changed from A〇 to A1, under data address 2 The data is changed from C0 to C1, and the above data are respectively stored in the storage unit-logic block of the logical disk. The logical disk is stored in blocks. The block size of all logical disks in the disk group is the same. The logical disk is equivalent to the partition in the non-LVM system, which is built on the disk group. It is a domain-based block device on which the text (10) system can be established. It is assumed that the physical size of the logical disk is the size of the logical disk. When the logical disk is created, the logical block and the physical block are established. One-to-one mapping relationship. By performing a pre-write copy operation, the snapshot created at the tl-day punctuality point first records the data A〇 and (3) to obtain the snapshot S11 content as shown in the "2a map", which includes The data A〇 and (3) to be written under the logical address corresponding to the two logical blocks. Figure 2b is a snapshot backup method according to the prior art, and a second snapshot is a schematic diagram of a point at the same time. Here, the second snapshot refers to a snapshot. Similarly, when the = point is established, a logical disk snapshot is created. At this time, the snapshot is still: empty t in the subsequent section _t2', that is, tf<t2<t2, there is a beacon address for the logical disk. 4 The data E0 is modified, from E〇 to £1, then the pre-write copy 1296776 is executed. (4)b目”_Reported 2 contents, exhausted by the snapshot and then woven: the data of the fast ^12 _ is also written The snapshot sn is written, and the content of the snapshot sir as shown in "2a" is as fast as the original snapshot Sli and the snapshot ^^^ gate: _ "3, the snapshot and logical magnetic created according to t3 The disk write operation of 'New Zealand===address 6' can be obtained as shown in the "2c picture", ", ", ", 2c picture" is a snapshot backup method according to the conventional technology, the third fast sii -r^S13 schematic diagram. Similarly, the 'snapshot S13 write also has a snapshot at the same time." "S12 write". At this time, the snapshot su, and the commits become snapshots S11... and the crimes shown in "%:" and "heart map" respectively. And "2b", at this time, the snapshot S11... contains the contents of the original snapshot, the decision, the S12, and the snapshot S13; and the snapshot S12 is the original snapshot S12 and the snapshot S13. Recorded data content. Incoming, knowing, the conventional snapshot backup method not only causes the total amount of backup data of the total parent to be written to multiple snapshots to gradually increase, thereby increasing the occupancy of all snapshots. The more times the snapshot is backed up, the larger the required capacity of the logical disk is "1 is larger". After the pre-write copy operation of the third snapshot, the snapshot S11 and each I reach 16ΜΒ. Line Description. The snapshot of the snapshot of the logical disk manager of the present invention is a snapshot of the snapshot point, and the "figure 5" is The snapshot processing method of the logical disk processor 1296776 of the present invention is a quick step flow chart. "A* Figure 4" is a schematic diagram of the logical magnetic monument of the present invention, and the snapshot method of the backup method is not in the middle. Assume that the logical disk is at the time point tr, and the logical address and data to be written, such as "1st and _. Μ", and the conventional technology is based on the 4th figure. The second snapshot of the quick resignation of the ducks; the flow: map. As shown in the figure below, when a plurality of snapshots have been created for the logical disk, when there is a writer for the logical disk at a certain time point, a snapshot created by the ^ day is found (step m), and In the snapshot of the last point in time, the data address of the data writer will be recorded in the j logical disk (step (4), and then the copy operation will be performed before the write, and the data address of the upcoming data writer will be found. Recording into this snapshot (step 1G3), after the snapshot recording, the data writing to the logical port ', the corresponding data address is performed (step 104). Obviously, if the logical disk only has a snapshot, the present invention The logical face manager snapshot management party knows that the effect of the snapshot backup technology is a marriage. The snapshot backup method of the snapshot management method of the logical disk management of the present invention can be combined with the "figure 4" and the conventional snapshot backup technology. The description is carried out. ^ Also assume that the logical disk at each time of the address and its corresponding original data ^ valley as shown in the data area 301 of "Fig. 3", and at a certain point in time ti on the current logical disk Data establishment (4) S21 At this time, no data fields are recorded in the snapshot. In the logical block with the bucking address of 〇 and 2, tl /, tl < tl / time point is executed 13 ^ 296776 Logic = two... The snapshot backup method of the present invention is needed The 1=Γ=Γ axis is divided into the initial created empty snapshot, so that the material with the address of 4 is entered into the factory. When there is a resource for the logical disk, the point t2 is fast and S22, and then the write is performed. The copy operation obtains the data in the corresponding time photo. The snapshot backup method according to the present invention is performed on the snapshot at the last time point. Similarly, at time t3 of the t3:=: one, that is, t2 - d ^~,...,S23, when there is a data η of 6 after ρ, there is a fast s23 change to the data address of the logical disk, and then the pre-write copy operation is performed to obtain the corresponding time point 『, * ' and according to this (4) The quick copy method, when the data in this Zaojia coffee is written, there is no snapshot of the fast time S2 $1, S23 to the last time point; 1 S22 write write, that is, the copy before the write only when the (four)' The size required for the snapshot s21 is 8 mb, which is only equivalent to the fineness and (4) the required size of the micro is 働, the phasor The size of the space is therefore much faster than that obtained by the prior art. ^^= The size of the occupied capacity is significantly reduced. The U above is a fast-forward one=: The invention is based on the snapshot backup method of the disk manager. Logical disk manager snapshot reading method. "Figure 6" is a flowchart of the snapshot reading step of the 14 I2%776 snapshot management method of the logical disk manager of the present invention. As shown in the figure, there is a logical disk. When one of the snapshot reads the request, a snapshot is selected according to the snapshot read request (step 1〇5), and then it is determined whether there is a data address of the read request in the selected snapshot (step 106), if any, Then directly reading one of the snapshot data corresponding to the request data address in the fast/photo (step 110); otherwise, determining whether the time point corresponding to the current snapshot is a last time point (step 107), and if yes, proceeding to the step 109. Read the ♦-original data corresponding to the corresponding request data address in the logical disk; if not the last time point, search backward for a plurality of snapshots later than the time point corresponding to the snapshot (step 108), and then determine the search. Fast Whether there is a data address of the read request in the photo (step 106), if yes, proceed to step 11〇, read the snapshot towel request data address lining-snapshot (four), and so on until the corresponding time point is A snapshot of the last point in time. The snapshot backup method of the logical disk manager of the present invention maintains snapshots of different time, and recovers through the snapshot reading method of the present logical disk manager when the logical disk is mishandled and the data is lost. . In the following, the snapshot reading method of the present invention, or the snapshot recovery method, will be described in conjunction with "Fig. 3; 1 and 帛 4". If the snapshot S22 exists - the current read request, the selection of the snapshot / the time point of the recovery in the disk is related to the time point, here, the selection of the snapshot micro means that the request face of the floppy disk needs to be performed on the t2 day. Read. If you want to read the snapshot data corresponding to the beetle address in I and S22 for 4 times, it can be seen from the figure that the data address 4 exists in the snapshot, so it is directly

15 1296776 Read according to the data E0. If the read request of S22 requires reading the snapshot data of the data address of 6, the content of the snapshot is not determined by the snapshot S22, and then it is determined whether the time point corresponding to the snapshot S22 is the last time point, as shown in the figure. There is a snapshot S23 at a time point of the snapshot S22, so the snapshot S23 is searched backwards, and it is judged whether or not the data address 6 exists in the snapshot S23. Since the data address 6 exists in the snapshot 兕3, the snapshot material GO under this data address is read. In another case, there is a read request for the data address 3 in the snapshot, and whether the data address exists in the snapshot micro or in the snapshot S23 of the snapshot, the data is directly read. Address 3 is the data content D() in the logical disk. That is, when the snapshot micro is required to be accessed, its content needs to be jointly presented by the snapshot S22 and the snapshot S23, thereby obtaining the correct result of the original content of the logical disk at the time t2, such as the data area 303 of the table 3 of "Fig. 3". Nothing. 'Calling, far from the snapshot reading method of the present invention, it is known that when accessing the snapshot milk, the content of the logical disk at the time of ti needs to be presented by snapshots S21, S22 and Qingqiu, as shown in Figure 3. In Table 3, the S23 can directly read the logical disk as shown in the data area 304 in t3. According to the logic track of the present invention, there is a __ between the grammar and the ββ, except for the snapshot obtained by the method of .n, which can be relied on by other snapshots, because the __ is caused by the __ =::::With:: 16 1296776

By snapshot merge mode, in the snapshot. It can be seen from the above description that the snapshot management method of the logical disk manager of the present invention can not only save the snapshot storage space, but also reduce the amount of data writing, and thus, the efficiency of the knowledge. In addition, the amount of data in the snapshot reading method and the amount of data in the conventional method are not good for the efficiency of reading the data - although the preferred embodiment of the present invention is as described above, It is not intended to use (10) to define the present invention, and it is within the spirit and scope of the present invention, and the patent protection of the present invention is defined as a part of the invention. Scales. ^ [Simple description of the diagram] Figure 1 is a schematic diagram of the data write address and its data content of the logical disk at different time points. The snapshot is in different time. The 2a picture is based on the snapshot backup method of the prior art. Schematic diagram of the point;

Schematic diagram of the point; 'The first snapshot is not at the same time. FIG. 2c is a snapshot of the snapshot method according to the prior art.

1296776 Intention; FIG. 4 is a schematic diagram of a snapshot backup method of a logical disk manager of the present invention at different time points; FIG. 5 is a snapshot backup/step flow chart of a snapshot management method of the logical disk manager of the present invention; And FIG. 6 is a flow chart of the snapshot read φ step of the snapshot management method of the logical disk manager of the present invention. [Main component symbol description] 10 Logic block area 20 Logic block area 30 Logic block area tf Time point t2 Time point t3 Time point S11 Snapshot

Slf snapshot

Sll^ Snapshot 18 1296776 S12 Snapshot S12 - Snapshot S13 Snapshot S21 Snapshot S22 Snapshot S23 Snapshot 30 Table 301 Data Area 302 Data Area 303 Data Area 304 Data Area Step 101 When a snapshot of a later time point is written, find the most step In the snapshot of the last time point, the data address of the upcoming data address is searched for. The step 103 of the stomach input is performed according to the write copy operation, and the data under the data write address is backed up into the snapshot. 19 1296776 Step 104 Performing the logic magnetic The write operation step 105 of the data address of the disc selects a snapshot step 106 according to the snapshot read request to determine whether there is a data address of one of the read requests in the snapshot. Is the time point corresponding to the snapshot in step 107 a last time point? Step 108: searching for a plurality of snapshots later than the corresponding time point of the snapshot. Step 109: reading the original data under the corresponding request data address in the logical disk. Step 110: reading one snapshot data corresponding to the requested data address in the snapshot.

20

Claims (1)

1296776 X. The scope of application for patents: 1 · · - The fast-track financial method of the magnetic disk drive, the financial law includes the following steps · When there is a request for information on a logical disk, the logical disk Build a plurality of snapshots corresponding to different days and points, find the data address of the logical disk to be written data; 'copy the month' to copy the last time point of the corresponding snapshot, the logic The disk is about to be performed (4) the data of the person's information is backed up; when there is a snapshot-reading request, it is determined whether there is a requested data address in one of the snapshots corresponding to the read request; The requested data address does not exist in the snapshot, and the corresponding time point of the snapshot is not - the most pie _ hour' _ after the rhyme is later than the (four) corresponding to the time (the fifth point) depends on the snapshot 'until the search The request data address and read one of the snapshot data under the request data address. For example, if the requesting data address exists in the snapshot, the snapshot data of the logical disk manager described in the claim 1 is directly read. 3. The snapshot management method of the logical disk manager as described in claim 1, wherein the request data address does not exist in the snapshot, and the corresponding time point of the snapshot is the last time When it is clicked, it reads the original data in the logical disk corresponding to one of the 21 Γ 296776 material addresses. 4 soil such as Shen kiss patent range! The logical disk manager described in the item is about to earn a bribe, and the deletion should be the address of the data. ===== Axis is in the photo. L is a logic. The law, the financial method includes the following steps: * Shao · In the case of a logical magnetic data request, the snapshot has a plurality of snapshots corresponding to different time_, to find the data to be written in the logical disk The data address; and the data under the data address to which the logical disk is to be written is backed up according to a snapshot of the last time point corresponding to the write copy operation. The snapshot management method of the logical disk management II described in claim 5 of the patent domain further includes the following steps: /presence__, when reading the request, whether there is a request in one of the snapshots corresponding to the read request The data address; and the request data address does not exist in the backup snapshot, and the corresponding time point of the snapshot is not 疋 - the last time point... then the backward looking for the corresponding time of the snapshot Snapshot, until the search for the data address, and read one of the snapshot data under the request data address. 22 1296776 7·If you apply for the Snapshot Management Method of the Logic Disk Manager described in item 6 of the scope of the patent application, the Snapshot Management Method of the Logic Manager is described in the Snapshot. At the time, the data of one of the request data addresses is directly read. ,, Shi Shen #Special H, the logical management of the logical disk manager described in the sixth item, ', towel (five) " paper,,, towel does not exist the request: #料 address, and the corresponding When the _ is the last time point, the original data in the logical disk corresponding to the requested data address is read. 9. The snapshot manager T' of the logical disk processor described in item 5 of the patent scope, wherein when a snapshot needs to be deleted from the plurality of snapshots, and the corresponding time point of the deletion is known If not—the earliest time _, the data under the corresponding data address in the deleted snapshot is added to its previous snapshot. twenty three