Search

CN-114995763-B - Storage system and storage method of storage volume

CN114995763BCN 114995763 BCN114995763 BCN 114995763BCN-114995763-B

Abstract

The invention relates to a storage system and a storage method of a storage volume, and belongs to the technical field of data storage. The storage volume is divided into a plurality of data units with the same size, each data unit is provided with corresponding metadata used for managing the attribute information of the storage volume, each metadata is stored in a set metadata management cluster, the metadata management cluster stores and manages the metadata corresponding to each data unit, and when the storage volume is mapped, the storage volume can be mapped into different storage pools according to the granularity of the data units, so that the storage volume is not limited to a certain storage pool. Therefore, the storage system of the invention can map the storage volume into any logic boundary or physical boundary, so that the space of the storage volume is independent and independent from any boundary, and the expansion of the storage pool or the data migration of the whole storage volume is avoided.

Inventors

  • YU QIANG
  • CAO YUZHONG
  • ZHOU WENMING

Assignees

  • 华瑞指数云(河南)科技有限公司

Dates

Publication Date
20260505
Application Date
20220531

Claims (10)

  1. 1. A storage system comprises a storage volume and a storage pool, wherein the storage pool comprises a logic pool and/or a physical pool, the storage system further comprises a metadata management cluster, the storage volume is divided into at least three data units, the data units are virtual space units according to a set size, each data unit is provided with corresponding metadata used for managing attribute information of the corresponding data unit in the storage volume, each metadata is stored in the metadata management cluster to achieve storage and management of the metadata, the storage volume is mapped into different storage pools with the data units as granularity to achieve storage of the storage volume in different storage pools, the storage volume is mapped into any logic boundary or physical boundary, the metadata management cluster is independent of the storage pool, the logic boundary refers to the logic pool, the physical boundary refers to the physical pool, when the data units are subjected to storage mapping, the corresponding storage pools are selected according to service data requirements of the data units, mapping is conducted when the service data requirements of the data units change, the corresponding data units are adjusted according to the service data requirements of the changed, the corresponding data units are adjusted according to the changed service data requirements, and the adjusted storage pool is mapped from the adjusted storage pool after the adjusted storage pool is mapped.
  2. 2. The memory system of claim 1, wherein the set size is 512KB-4MB.
  3. 3. A storage system according to claim 1 or 2, wherein when a storage pool mapped with data units fails, data in the mapped data units in the storage pool is migrated to the other storage pool.
  4. 4. The storage system of claim 1, wherein when a storage pool capacity mapped by a data unit is full, other data units are no longer mapped to the storage pool.
  5. 5. The storage system of claim 1, wherein when traffic data of a data unit is changed from hot spot data requiring frequent reading and writing to normal data requiring no frequent reading and writing, the data unit is mapped from a storage pool with fast storage speed to a storage pool with normal storage speed.
  6. 6. A storage method of a storage volume, the storage method comprising the steps of: 1) Dividing the storage volume into at least three data units according to the set size, wherein the data units are virtual space units according to the set size; 2) Setting corresponding metadata for each data unit, and storing each metadata in a metadata management cluster, wherein the metadata is used for managing attribute information of the corresponding data unit in a storage volume; 3) Mapping the storage volumes into different storage pools by taking data units as granularity, so as to realize mapping the storage volumes according to the granularity of the data units, enabling the storage volumes to be mapped into any logic boundary or physical boundary, wherein the storage pools comprise logic pools and/or physical pools, the metadata management clusters are independent of the storage pools, the logic boundary refers to the logic pools, the physical boundary refers to the physical pools, when the data units are subjected to storage pool mapping, the corresponding storage pools are selected according to the service data requirements of the data units to be mapped, when the service data requirements of the data units are changed, the storage pools mapped by the corresponding data units are adjusted according to the changed requirements, and the data of the data units are transferred from the storage pools mapped before adjustment to the storage pools mapped after adjustment.
  7. 7. The method of claim 6, wherein the set size is 512KB-4MB.
  8. 8. The storage method of claim 6 or 7, wherein when a storage pool mapped by a data unit fails, data in the mapped data unit in the storage pool is migrated to another storage pool.
  9. 9. The storage method of claim 6, wherein when the storage pool capacity mapped by a data unit is full, other data units are no longer mapped to the storage pool.
  10. 10. The method according to claim 6, wherein when there is traffic data of a data unit changed from hot spot data requiring frequent reading and writing to normal data requiring no frequent reading and writing, the data unit is mapped from a storage pool having a fast storage speed to a storage pool having a general storage speed.

Description

Storage system and storage method of storage volume Technical Field The invention relates to a storage system and a storage method of a storage volume, and belongs to the technical field of data storage. Background The storage volume is specifically a block storage volume, which is a logical space unit of a block storage system used by a user, and generally corresponds to a virtual disk in the user system. A block storage volume in current block storage product architectures is assigned to a logical boundary or within a physical boundary and does not span this boundary dimension. Logical boundaries refer to a range of redundancy policies, also called a logical pool, for data reliability, whether conventional storage or distributed storage, where a set of redundancy policy algorithms exist to guarantee the reliability of multiple storages of data, such as RAID techniques, such as replica techniques, etc. The scope of the redundancy strategy is a logic space only in the protection space formed by the redundancy strategy, and the data in the scope is protected by the only one redundancy strategy in the scope. The physical boundary refers to the concept of a storage pool or a storage cluster of storage, and is collectively referred to as a physical pool, and whether conventional storage or distributed storage, there is one physical boundary that is a storage pool or a storage cluster, for a conventional block storage system, is an aggregate of all disks that form a single redundancy policy corresponding to a block storage volume, while for distributed storage, the range is wider, and not only an aggregate of disks but also an aggregate of a plurality of servers that form a distributed storage cluster. Since volumes in conventional storage architectures (particularly block storage architectures) are a subset of the space of logical pools, physical pools, this also results in users using block storage volumes being limited to being within the boundaries of these two dimensions. The volume is a logical space built on the upper layer of the storage system, and has certain overstock characteristics (namely, the effective space of the volume is larger than the space provided by actual physics, for example, the effective space of the volume is defined as 10T, and the storage space of a physical pool is 5T), so when the space of the logical pool and the physical pool is insufficient, as the space of the volume is limited by the boundaries of the three dimensions, new data written into the volume by a user cannot be written in, and the new data can be only solved by continuous expansion of the bottom space, but the expansion can lead to rebalancing of the whole storage cluster, and the use experience of other storage volumes in the whole boundary can be influenced. If the capacity expansion occurs that resources are tense and capacity expansion cannot be performed in time, the service cannot be used, and further the continuity of the service is affected. In addition, if the redundancy strategy in the boundary is in a state of fault degradation for a long time and a fault which cannot be repaired in a short time occurs, if any one level of fault occurs in addition, the whole data is required to be moved to other logic pools and physical pools, and the movement of a large amount of data will affect the continuity of the service. In order to solve the problems, a cross-pool volume migration technology is presented, namely, a set of software or a set of program is added outside a storage system, and the storage volume can be copied from one logical pool or physical pool and written into the other logical pool or physical pool, so that the limitation caused by the prior inability to cross is broken, and the two-point constraint problem can be solved to a certain extent. The scheme is realized outside the band of the storage system, an additional splicing function is needed for the storage system, huge data relocation cost is brought, and the relocation process is to relocate the whole block storage volume. For example, if the logical pool and the physical pool are insufficient in space or degraded and cannot be recovered in a short time, an online volume migration function of some volumes needs to be started in an emergency mode to enable more available space to be moved or to migrate volumes in a fault range to a normal cluster range, and a large number of background IOs are generated in the migration process due to the fact that the whole volume is required to be migrated in the migration process, the background IOs can influence business IOs of a foreground, the whole business IO fluctuates, and the recovery can be achieved only after the migration is completed. Therefore, the current storage volume migration technology does not solve the problem from architecture at all due to the problem of architecture, only solves the problem from out-of-band mode, is limited by the capacity of the storage pool, and still has a large