Shulou(Shulou.com)06/02 Report--

[detailed explanation of IBM principle]

IBM_V7000 (full name IBM Storwize V7000) is a new generation of midrange storage system launched by IBM. Although it is located in the midrange, Storwize V7000 provides the powerful storage management functions that high-end storage has in the past. Its common models are IBM Storwize V3700 IBM Storwize V5000 and IBM Storwize V7000. Its underlying storage structure supports RIAD 0Grand raid 10Gen raid 5 and RAID 6. The upper volumes support normal volumes, compact mode volumes, mirrored mode volumes, and compact mirror mode volumes. Although V7000 does a good job in the overall storage structure, some physical failures or other operations can cause damage to volumes or storage, so the data recovery technology for series storage has the opportunity to show its ability.

[configure IBM_V7000]

1. Use the administrative IP to connect to IBM_V7000, and enter the user name (default: superuser) and password (password: passw0rd).

2. There is no configuration by default. Mdisk, pool and volume need to be configured first, and then the host mapping volume is added.

3. The types of creating Mdisk,Mdisk are RAID0,RAID0,RAID5 and RAID6.

4. Create a pool, which can divide multiple Mdisk into one pool.

5. Create a volume. The volume is configured on the basis of the pool. The types of volumes are generic, automatic thinning, mirroring and thin mirroring.

6. Create hosts and map volumes. The types of hosts are fibre Channel hosts and iscsi hosts.

At this point, the general process of the whole configuration is over, but we do not know how the logical volumes assigned to the host are actually distributed on disk. So how are they distributed? Explain the structure and principle of the following in detail, its structure is very similar to the storage products of HP Lefthand series.

[structure and principle]

In fact, the underlying principle of IBM_V7000 is not complex, the whole storage structure is divided into four layers.

The first layer: the physical hard disk is the place where the data is actually stored.

The second layer: IBM_V7000 named Mdisk, which is actually RAID, is a collection of multiple physical disks.

The third layer, the pool, is the combination of multiple Mdisk into a large logical container.

Layer 4: volumes, space allocated from the pool, user-oriented storage units, volumes cannot cross pools.

The structure diagram is as follows:

From the overall storage structure, the disk is the final place where the data is stored. The so-called Mdisk, pools and volumes all virtualize physical disks. In the physical disk layer, the data is stored in small blocks (Block), N multiple disks constitute a Mdisk, and the data stored in the Mdisk will be divided into N multiple Block evenly distributed on all disks. At the Mdisk tier, data is stored in Section, and multiple Mdisk form a pool. Volumes created in the pool are divided into several segments and placed in different Mdisk, and different volume types are distributed in different ways in the pool, but in the end, segments are stored in Midsk.

The whole stored procedure is that the user stores the data in the volume, and the volume will be divided into several segments and distributed in different Mdisk, while the Mdisk will divide the segments into several blocks and distribute them on different disks. The final data is all distributed on different disks in blocks.

[data recovery case]

1. Storage architecture

Storage model: IBM_V7000

Number of disks: 24 600g SAS disks

Number of Mdisk: 2 Mdisk, both RAID 5

Number of volumes: 2 x 2T general mode, 1 x 3T compact mode.

2. Cause of failure

Several disks in Mdisk are offline due to disk aging, resulting in the unavailability of Mdisk and the inaccessibility of upper-layer volumes. Because only one global hot standby is set, the manager does not change the disk in time after the disk is dropped, which leads to the whole failure.

3. Solution

Mirror all disks first, and then analyze which disks are a set of Mdisk. Find out if you belong to the same group of Mdiskd disks, and then analyze if there is a disk drop. If there is an offline disk, remove it from the Mdisk group. Generate Mdisk, and then analyze the structure between Mdisk. Generate a pool, and then analyze the structure of the volume, because the type of different volumes is different, so the storage structure is also different. After analyzing the structure of the volume, you can generate data for each volume.

4. Data recovery result

Since only a few disks in Mdisk have gone offline, no other operations have been done. Therefore, the whole data recovery is very complete, and the user acceptance data is correct.

Due to the thorough study of the underlying structure of IBM V7000 series storage, almost all the failures of this series of storage can be saved. However, there is a situation where the data is irreparable, that is, all disks have been rebuilt and initialized. Because after the creation of the Mdisk, the system will initialize the Mdisk, that is, zero. The figure below is as follows

Therefore, if the storage is rebuilt and initialized, there is little hope of recovering the data in this case.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.