DISTRIBUTED STORAGE SYSTEM AND CONTROL METHOD FOR DISTRIBUTED STORAGE SYSTEM

• US 20180074746A1
• Filed: 03/16/2015
• Published: 03/15/2018
• Est. Priority Date: 03/16/2015
• Status: Active Grant

• Current Assignee: Hitachi, Ltd.
• Original Assignee: Hitachi, Ltd.
• Inventors: WATANABE, Yasuo, AKUTSU, Hiroaki
• Granted Patent: US 10,346,075 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 12/10 Address translation
• CPC Class Codes: G06F 2212/1044 Space efficiency improvement
• CPC Class Codes: G06F 2212/152 Virtualized environment, e.g. logically partitioned system
• CPC Class Codes: G06F 2212/154 Networked environment
• CPC Class Codes: G06F 3/06 Digital input from, or digital output to, record carriers , e.g. RAID, emulated record carriers or networked record carriers
• CPC Class Codes: G06F 3/0608 Saving storage space on storage systems
• CPC Class Codes: G06F 3/061 Improving I/O performance
• CPC Class Codes: G06F 3/0641 De-duplication techniques
• CPC Class Codes: G06F 3/0665 at area level, e.g. provisioning of virtual or logical volumes
• CPC Class Codes: G06F 3/067 Distributed or networked storage systems, e.g. storage area networks [SAN], network attached storage [NAS]

Abstract

Regarding a distributed storage system including a plurality of nodes, a first node among the plurality of nodes judges whether the same data as first data, which is written to a first virtual partial area managed by the first node from among a plurality of virtual partial areas, exists in the virtual partial area managed by another node among the plurality of nodes; when the same data as the first data exists in the other node, the first node executes inter-node deduplication for changing allocation of either one of logical partial areas for the first virtual partial area or the virtual partial area of the other node to which the same data is written, to the other logical partial area; and when I/O load on the first node after execution of the inter-node deduplication of the first virtual partial area and the predicted value is less than a first threshold, the first node executes the inter-node deduplication of a second virtual partial area managed by the first node from among the plurality of virtual partial areas obtained by dividing the virtual storage area.

Claims

1. A distributed storage system including a plurality of nodes, which are connected to each other and each of which includes a storage medium, for controlling storage of data in logical storage areas composed of the storage medium,

wherein the plurality of nodes:

provide a host system with a virtual storage area across the logical storage areas controlled by the respective nodes;

manage allocation of a plurality of virtual partial areas obtained by dividing the virtual storage area and a plurality of partial logical storage areas obtained by dividing the logical storage area; and

distribute data written to the virtual storage area and store the data in the logical storage areas;

wherein a first node among the plurality of nodes judges whether the same data as first data, which is written to a first virtual partial area managed by the first node from among the plurality of virtual partial areas, exists in the virtual partial area managed by another node among the plurality of nodes;

wherein when the same data as the first data exists in the other node, the first node executes inter-node deduplication for changing allocation of either one of logical partial areas for the first virtual partial area or the virtual partial area of the other node to which the same data is written, to the other logical partial area; and

wherein when the first node calculates a predicted value of I/O load on the first node after execution of the inter-node deduplication of the first virtual partial area and the predicted value satisfies a predetermined condition, the first node executes the inter-node deduplication of a second virtual partial area managed by the first node from among the plurality of virtual partial areas obtained by dividing the virtual storage area.

2. The distributed storage system according to claim 1,

wherein the I/O load is at least one of inter-node reading network load, average response time, and average throughput;

and

wherein the predetermined condition is satisfied when the inter-node reading network load is less than a first threshold, when the average response time is less than a second threshold, or when the average throughput is more than a third threshold.

3. The distributed storage system according to claim 1,

wherein the first node executes the inter-node deduplication on data stored in the first node sequentially from data in a virtual partial area with low I/O frequency.

4. The distributed storage system according to claim 1,

wherein when the first node judges whether or not the same data exists between virtual partial areas managed by the first node from among the plurality of virtual partial areas and when the same data exists, the first node executes node deduplication for aggregating allocations of the logical partial areas to the virtual partial area into one logical partial area and then executes the inter-node deduplication.

5. The distributed storage system according to claim 1,

wherein a total sum of a deduplicated number of the first data stored in the first node and a deduplicated number of second data stored in a second node is less than a second threshold, the first node executes the inter-node deduplication of the first data and the second data.

6. The distributed storage system according to claim 1,

wherein when the first node receives a read request from the host system to the first virtual partial area and when the logical partial area of the other node is allocated to the first virtual partial area, the first node acquires data requested by the read request from the other node and transmits the acquired data to the host system.

7. The distributed storage system according to claim 6,

wherein when the first node receives a first write request for a third virtual partial area of the virtual storage area, the first node allocates a third logical partial area from among the plurality of logical partial areas and writes data requested by the first write request to the third logical partial area;

and

wherein when the first node receives a second write request for a fourth virtual partial area, the first node changes allocation from the first partial logical storage area to a fourth logical partial area from among the plurality of logical partial areas and writes data requested by the second write request to the fourth logical partial area.

8. A control method for a distributed storage system including a plurality of nodes, which are connected to each other and each of which includes a storage medium, for controlling storage of data in logical storage areas composed of the storage medium, the control method for the distributed storage system, comprising:

providing a host system with a virtual storage area across the logical storage areas controlled by the plurality of nodes respectively, managing allocation of a plurality of virtual partial areas obtained by dividing the virtual storage area and a plurality of partial logical storage areas obtained by dividing the logical storage area, and

distributing data written to the virtual storage area and storing the data in the logical storage areas;

judging whether the same data as first data, which is written to a first virtual partial area managed by a first node from among the plurality of virtual partial areas, exists in the virtual partial area managed by another node among the plurality of nodes;

executing inter-node deduplication, when the same data as the first data exists in the other node, for changing allocation of either one of logical partial areas for the first virtual partial area or the virtual partial area of the other node to which the same data is written, to the other logical partial area; and

executing the inter-node deduplication of a second virtual partial area managed by the first node from among the plurality of virtual partial areas obtained by dividing the virtual storage area when a predicted value of I/O load on the first node after execution of the inter-node deduplication of the first virtual partial area is calculated and the predicted value satisfies a predetermined condition.

9. The control method for the distributed storage system according to claim 8,

wherein the I/O load is at least one of inter-node reading network load, average response time, and average throughput;

and

wherein the predetermined condition is satisfied when the inter-node reading network load is less than a first threshold, when the average response time is less than a second threshold, or when the average throughput is more than a third threshold.

10. The control method for the distributed storage system according to claim 8,

wherein the inter-node deduplication is executed on data stored in the first node sequentially from data in a virtual partial area with low I/O frequency.

11. The control method for the distributed storage system according to claim 8,

wherein when whether or not the same data exists between virtual partial areas managed by the first node from among the plurality of virtual partial areas is judged and when the same data exists, node deduplication is executed for aggregating allocations of the logical partial areas to the virtual partial area into one logical partial area and then the inter-node deduplication is executed.

12. The control method for the distributed storage system according to claim 8,

wherein a total sum of a deduplicated number of the first data stored in the first node and a deduplicated number of second data stored in a second node is less than a second threshold, the inter-node deduplication of the first data and the second data is executed.

13. A distributed storage system including a first node connected to one or more nodes, each of which has a storage medium, for controlling storage of data in a logical storage area composed of the storage medium, the first node and the one or more nodes constituting the distributed storage system,

wherein the first node:

provides a host system with a virtual storage area across the logical storage area controlled by the one or more nodes and the logical storage area controlled by the first node;

manages allocation of a plurality of virtual partial areas obtained by dividing the virtual storage area and a plurality of partial logical storage areas obtained by dividing the logical storage area; and

distributes data written to the virtual storage area with the one or more nodes and stores the data in the logical storage area;

wherein the first node judges whether the same data as first data, which is written to a first virtual partial area managed by the first node from among the plurality of virtual partial areas, exists in the virtual partial area managed by another node among the plurality of nodes;

wherein when the same data as the first data exists in the other node, the first node executes inter-node deduplication for changing allocation of either one of logical partial areas for the first virtual partial area or the virtual partial area of the other node to which the same data is written, to the other logical partial area; and

wherein when the first node calculates a predicted value of I/O load on the first node after execution of the inter-node deduplication of the first virtual partial area and the predicted value satisfies a predetermined condition, the first node executes the inter-node deduplication of a second virtual partial area managed by the first node from among the plurality of virtual partial areas obtained by dividing the virtual storage area.