Distributed file system

• US 8,943,022 B2
• Filed: 03/26/2012
• Issued: 01/27/2015
• Est. Priority Date: 03/29/2011
• Status: Active Grant

• Current Assignee: NEC Corporation
• Original Assignee: NEC Corporation
• Inventors: Takemoto, Taira
• Primary Examiner(s): Beausoliel, Jr., Robert
• Assistant Examiner(s): BROMELL, ALEXANDRIA Y
• Application Number:
  US13/430,575
• Publication Number:
  US 20120254110A1
• Time in Patent Office: 1,037 Days
• Field of Search: 707/624
• US Class Current: 707/624
• CPC Class Codes: G06F 16/184 implemented as replicated file system

Abstract

A computer node connected with an upstream computer node and a downstream computer node so as to be capable of communicating therewith has a replication file system of a master file system, a storing unit, and a controlling unit. The controlling unit receives incremental data showing a changed part of the master file system from the upstream computer node, stores the incremental data into the storing unit, reflects the incremental data on the file system, and transfers the incremental data to the downstream computer node. Moreover, in a case that incremental data for plural times having received from the upstream computer node and having not transferred to the downstream computer node yet are accumulated in the storing unit, the controlling unit merges, with last received incremental data among the incremental data for plural times, the rest of the incremental data for plural times.

Claims

1. A computer system, comprising:

a first computer node comprising a first file system, wherein the file system is stored on a non-transitory computer readable recordable medium;

a second computer node connected to the first computer node so as to be capable of communicating therewith and comprising a second file system as a replica of the first file system; and

a third computer node connected to the second computer node so as to be capable of communicating therewith and comprising a third file system as a replica of the first file system,

wherein the first computer node includes an incremental data generating unit configured to generate a set of changed blocks from a previous incremental data generation moment among blocks of the first file system as incremental data, a first storing unit, and a first controlling unit configured to store the incremental data generated by the incremental data generating unit into the first storing unit and transmit to the second computer node,

wherein the incremental data generating unit acquires a snapshot that includes an image of the first file system for every change and addition or at regular intervals, and calculates a difference between a currently acquired snapshot and a previously acquired snapshot as the incremental data,

wherein the second computer node includes a second storing unit, and a second controlling unit configured to store the incremental data received from the first computer node into the second storing unit, reflect on the second file system, and transmit to the third computer node,

wherein the third computer node includes a third storing unit, and a third controlling unit configured to store the incremental data received from the second computer node into the third storing unit and reflect on the third file system, and

wherein, when plural incremental data received from the first computer node are accumulated in the second storing unit, before transmitting the plural incremental data to the third computer node, the second controlling unit merges plural sets of changed blocks of the plural incremental data into merged one incremental data, and then the second controlling unit transmits the merged one incremental data, rather than each piece of the plural incremental data received from the first computer node, to the third computer node.

2. The computer system according to claim 1, wherein the second controlling unit deletes a remaining incremental data after the merging from the second storing unit.

3. The computer system according to claim 2, wherein the second controlling unit transmits a request for deleting the incremental data deleted from the second storing unit to the first controlling unit, and

wherein the first controlling unit deletes the incremental data relating to the deletion request from the first storing unit.

4. The computer system according to claim 1, wherein the incremental data generating unit generates the set of changed blocks from the previous incremental data generation moment among blocks configuring the first file system, as the incremental data.

5. The computer system according to claim 4, wherein in a process of the merging, the second controlling unit generates the incremental data that includes a block included in only one of the plural incremental data and, in a case that a block is included in each of two or more incremental data, includes the block included in the last received incremental data of the two or more incremental data.

6. The computer system according to claim 1, wherein the second controlling unit merges the plural sets of changed blocks of the plural incremental data by merging a last received incremental data among the plural incremental data with a remaining of the plural incremental data, and deleting the remaining of the plural incremental data from the second storing unit.

7. The computer system according to claim 1, wherein the second controlling unit merges changed blocks between a changed source file at a certain time and a corresponding destination file into changed blocks between the changed source file at a later time and the corresponding destination file.

8. A multiplexing method executed by a distributed file system which includes a first computer node comprising a first file system, a second computer node connected to the first computer node so as to be capable of communicating therewith and comprising a second file system as a replica of the first file system, and a third computer node connected to the second computer node so as to be capable of communicating therewith and comprising a third file system as a replica of the first file system, and in which the first computer node includes an incremental data generating unit, a first storing unit and a first controlling unit, the second computer node includes a second storing unit and a second controlling unit, and the third computer node includes a third storing unit and a third controlling unit,

the multiplexing method in the distributed file system, comprising:

generating a set of changed blocks from a previous incremental data generation moment among blocks of the first file system as incremental data, by the incremental data generating unit of the first computer node;

storing the incremental data generated by the incremental data generating unit into the first storing unit and transmitting to the second computer node, by the first controlling unit of the first computer node;

wherein, in the generation of the incremental data by the incremental data generating unit, a snapshot that includes an image of the first file system is acquired for every change and addition or at regular intervals and a difference between a currently acquired snapshot and a previously acquired snapshot is calculated as the incremental data,

storing the incremental data received from the first computer node into the second storing unit, reflecting on the second file system, and transmitting to the third computer node, by the second controlling unit of the second computer node;

storing the incremental data received from the second computer node into the third storing unit and reflecting on the third file system, by the third controlling unit of the third computer node; and

when plural incremental data received from the first computer node are accumulated in the second storing unit, before transmitting the plural incremental data to the third computer node, merging plural sets of changed blocks of the plural incremental data into merged one incremental data, and then the second controlling unit transmits the merged one incremental data, rather than each piece of the plural incremental data received from the first computer node, to the third computer node.

9. The multiplexing method in the distributed file system according to claim 8, further comprising deleting a remaining incremental data after the merging from the second storing unit, by the second controlling unit of the second computer node.

10. The multiplexing method in the distributed file system according to claim 9, further comprising:

transmitting a request for deleting the incremental data deleted from the second storing unit to the first controlling unit, by the second controlling unit; and

deleting the incremental data relating to the deletion request from the first storing unit, by the first controlling unit.

11. The multiplexing method in the distributed file system according to claim 8, wherein in the generation of the incremental data by the incremental data generating unit, a set of blocks changed from a previous incremental data generation moment among blocks configuring the first file system is generated as the incremental data.

12. The multiplexing method in the distributed file system according to claim 11, wherein, in the merging by the second controlling unit, the incremental data that includes a block included in only one of the plural incremental data and, in a case that a block is included in each of two or more incremental data, includes the block included in the last received incremental data is generated.

13. The multiplexing method in the distributed file system according to claim 8, wherein the second controlling unit merges changed blocks between a changed source file at a certain time and a corresponding destination file into changed blocks between the changed source file at a later time and the corresponding destination file.

14. A computer node connected to an upstream computer node and a downstream computer node so as to be capable of communicating therewith, and including a replication file system that is a replica of a master file system, a storing unit, and a controlling unit, stored on a non-transitory computer readable recordable medium,

wherein the controlling unit receives incremental data including a set of changed blocks from a previous incremental data generation moment among blocks of the master file system from the upstream computer node, stores into the storing unit, reflects on the replication file system, transmits to the downstream computer node and, when plural incremental data received from the upstream computer node are accumulated in the storing unit, before transmitting the plural incremental data to the downstream computer node, the controlling unit merges plural sets of changed blocks of the plural incremental data into merged one incremental data and then the controlling unit transmits the merged one incremental data, rather than each piece of the plural incremental data received from the upstream computer node, to the downstream computer node,

wherein a snapshot is created that includes an image of the upstream file system for every change and addition or at regular intervals, and calculates a difference between a currently acquired snapshot and a previously acquired snapshot as the incremental data.

15. The computer node according to claim 14, wherein the controlling unit deletes a remaining incremental data after the merging from the second storing unit.

16. The computer node according to claim 15, wherein the controlling unit transmits a request for deleting the incremental data deleted from the storing unit to the upstream controlling unit.

17. The computer node according to claim 16, wherein, in a process of merging a remaining incremental data other than the incremental data having been received last among the incremental data for plural times with the incremental data having been received last among the incremental data for plural times, the controlling unit generates incremental data that includes a block included in only one of the incremental data for plural times and, in a case that a block is included in each of two or more incremental data, includes the block included in the last received incremental data.

18. The computer node according to claim 14, wherein the controlling unit merges changed blocks between a changed source file at a certain time and a corresponding destination file into changed blocks between the changed source file at a later time and the corresponding destination file.