Methods and apparatus for efficient container location database snapshot operation
The efficiency of a container location database (“CLDB”) that is used for retrieving location information of data containers stored on nodes in a cluster is improved by decreasing the number of key-value store transactions performed during snapshot creation and deletion, thereby decreasing both disk I/O and computing cycles. The CLDB is made more scalable by improving the deletion and creation time of snapshots, allowing one to create and delete more snapshots during a given interval of time. CLDB resilience is improved by decreasing the time required to resync a CLDB node. CLDB data is reduced to decrease IO operations, improve resync times, and to cache most of the data in memory.
1. A computer-implemented method comprising:
generating a volume snapshot for a volume in a distributed file system, the volume snapshot including a plurality of snapshot containers, each of the plurality of snapshot containers having a respective snapshot container identifier;
storing the snapshot container identifiers for the plurality of snapshot containers in one or more blocks, a first block of the one or more blocks including a contiguous sequence of multiple snapshot container identifiers for respective snapshot containers of the plurality of snapshot containers; and
storing information to specify the plurality of snapshot containers, wherein the information includes an identifier for the volume snapshot, a first snapshot container identifier at a beginning of the contiguous sequence of multiple snapshot container identifiers in the first block, and length information specifying a count of a number of snapshot container identifiers in the contiguous sequence of multiple snapshot container identifiers.
2. The computer-implemented method of claim 1 , wherein storing the information includes storing one or more key-value pairs in a key-value store of a container location database (CLDB).
3. The computer-implemented method of claim 1 , further comprising:
obtaining from the information a first set of information for the first block including the contiguous sequence of multiple snapshot container identifiers; and
converting the first set of information into a list of snapshot container identifiers for snapshot containers that are part of the volume snapshot.
4. The computer-implemented method of claim 1 , wherein the information specifies the plurality of snapshot containers as a table of entries, each entry representing a respective block of the one or more blocks.
5. The computer-implemented method of claim 1 , wherein the information specifies the plurality of snapshot containers in a single data structure.
6. The computer-implemented method of claim 1 , further comprising:
receiving a request for a current size of a volume snapshot container for the volume snapshot; and
computing the current size of the volume snapshot container based on a size of each of the plurality of snapshot containers and on updates received when snapshots are deleted.
7. The computer-implemented method of claim 1 , further comprising:
associating each respective snapshot container identifier of the snapshot container identifiers for the volume snapshot with an identifier of a corresponding read-write container, wherein a snapshot container identified by the respective snapshot container identifier is co-resident with the corresponding read-write container in a storage pool;
receiving a request for a location of a first snapshot container, the request including a given snapshot container identifier for the first snapshot container; and
determining the location of the first snapshot container based on a location of a first read-write container corresponding to the first snapshot container, the first read-write container being identified based on association of the given snapshot container identifier with an identifier of the first read-write container.
8. A non-transitory storage medium comprising instructions that upon execution cause one or more processors to:
generate a volume snapshot for a volume in a distributed file system, the volume snapshot including a plurality of snapshot containers, each of the plurality of snapshot containers having a respective snapshot container identifier;
store the snapshot container identifiers for the plurality of snapshot containers in one or more blocks, a first block of the one or more blocks including a contiguous sequence of multiple snapshot container identifiers for respective snapshot containers of the plurality of snapshot containers; and
store information to specify the plurality of snapshot containers, wherein the information includes an identifier for the volume snapshot, a first snapshot container identifier at a beginning of the contiguous sequence of multiple snapshot container identifiers in the first block, and length information specifying a count of a number of snapshot container identifiers in the contiguous sequence of multiple snapshot container identifiers.
9. The non-transitory storage medium of claim 8 , wherein the information includes one or more key-value pairs stored in a key-value store of a container location database (CLDB).
10. The non-transitory storage medium of claim 8 , wherein the instructions upon execution cause the one or more processors further to:
obtain from the information a first set of information for the first block including the contiguous sequence of multiple snapshot container identifiers; and
convert the first set information into a list of snapshot container identifiers for snapshot containers that are part of the volume snapshot.
11. The non-transitory storage medium of claim 8 , wherein the information comprises a table of entries, each entry representing a respective block of the one or more blocks.
12. The non-transitory storage medium of claim 8 , wherein the first snapshot container identifier at the beginning of the contiguous sequence of multiple snapshot container identifiers and the length information together specify an entire content of the first block.
13. The non-transitory storage medium of claim 8 , wherein the instructions upon execution cause the one or more processors further to:
receive a request for a current size of a volume snapshot container for the volume snapshot; and
compute the current size of the volume snapshot container based on a size of each of the plurality of snapshot containers and on updates received when snapshots are deleted.
14. The non-transitory storage medium of claim 8 , wherein of the instructions upon execution cause the one or more processors further to:
associate each respective snapshot container identifier of the snapshot container identifiers for the volume snapshot with an identifier of a corresponding read-write container, wherein a snapshot container is co-resident with the corresponding read-write container in a storage pool;
receive a request for a location of a first snapshot container, the request including a given snapshot container identifier for the first snapshot container; and
determine the location of the first snapshot container based on a location of a first read-write container corresponding to the first snapshot container, the first read-write container being identified based on association of the given snapshot container identifier with an identifier of the first read-write container.
15. A system comprising:
one or more cluster nodes, each cluster node including a storage pool; and
a container location database (CLDB) coupled with the one or more cluster nodes, the CLDB to maintain location information for data containers stored in the storage pools of the cluster nodes;
a processor; and
a non-transitory storage medium storing instructions executable on the processor to:
generate a volume snapshot for a volume of data in the one or more cluster nodes, the volume snapshot including a plurality of snapshot containers, each of the plurality of snapshot containers having a respective snapshot container identifier,
store the snapshot container identifiers for the plurality of snapshot containers in one or more blocks, a first block of the one or more blocks including a contiguous sequence of multiple snapshot container identifiers for respective snapshot containers of the plurality of snapshot containers, and
store a key-value pair in a key-value store of the CLDB to specify the plurality of snapshot containers, wherein the key-value pair includes an identifier for the volume snapshot, a starting snapshot container identifier at a beginning of the contiguous sequence of multiple snapshot container identifiers in the first block, and length information specifying a count of a number of snapshot container identifiers in the contiguous sequence of multiple snapshot container identifiers.
16. The system of claim 15 , wherein the instructions are executable on the processor to:
obtain the key-value pair from the key-value store; and
convert information in the key-value pair into a list of snapshot container identifiers for snapshot containers that are part of the volume snapshot.
17. The system of claim 15 , wherein the starting snapshot container identifier at the beginning of the contiguous sequence of multiple snapshot container identifiers and the length information together specify an entire content of the first block.
18. The system of claim 15 , wherein the key-value pair comprises:
a key that includes the identifier for the volume snapshot and the starting snapshot container identifier, and a value that includes the length information, or
a key that includes the identifier for the volume snapshot, and a value that includes the starting snapshot container identifier and the length information.
19. The system of claim 15 , wherein the instructions are executable on the processor to:
receive a request for a current size of a volume snapshot container for the volume snapshot; and
compute the current size of the volume snapshot container based on a size of each of the plurality of snapshot containers and on updates received when snapshots are deleted.
20. The system of claim 15 , wherein the instructions are executable on the processor to:
associate each respective snapshot container identifier of the snapshot container identifiers for the volume snapshot with an identifier of a corresponding read-write container, wherein a snapshot container identified by the respective snapshot container identifier is co-resident with the corresponding read-write container in a storage pool;
receive a request for a location of a first snapshot container, the request including a given snapshot container identifier for the first snapshot container; and
determine the location of the first snapshot container based on a location of a first read-write container corresponding to the first snapshot container, the first read-write container being identified based on association of the given snapshot container identifier with an identifier of the first read-write container.