Generating merge-friendly sequential IO patterns in shared logger page descriptor tiers
Techniques for generating merge-friendly sequential IO patterns in shared logger page descriptor (PD) tiers. The techniques can be employed in an active-active clustered system that includes a primary storage node, a secondary storage node, a logger shared between the primary and secondary storage nodes, and a storage device such as a non-volatile memory (NVM) device. The techniques can include allocating at least a first trunk and a second trunk in a shared PD tier of the logger, sequentially writing PD metadata and/or data by the primary storage node and the secondary storage node to the first trunk and the second trunk, respectively, merging the PD metadata/data units sequentially written to each respective trunk together to obtain PD metadata/data having an increased size up to the size of the respective trunks, and storing the merged PD metadata/data to the NVM device.
1. A method of generating merge-friendly sequential input/output (IO) patterns in a shared page descriptor (PD) tier, comprising:
allocating at least a first trunk and a second trunk in a shared PD tier of a logger, wherein the first trunk and the second trunk are different trunks;
sequentially writing first PD metadata and/or data by a primary storage node IQ the first trunk, and sequentially writing second PD metadata and/or data by a secondary storage node to the second trunk;
merging the first PD metadata and/or data sequentially written to the first trunk to obtain first merged PD metadata and/or data, and merging the second PD metadata and/or data sequentially written to the second trunk to obtain second merged PD metadata and/or data; and
storing the first merged PD metadata and/or data and the second merged PD metadata and/or data to a non-volatile memory (NVM) device.
2. The method of claim 1 , wherein the first trunk has an associated head pointer and an associated tail pointer, and wherein sequentially writing the first PD metadata and/or data to the first trunk includes writing the first PD metadata and/or data to a head of the first trunk.
3. The method of claim 2 , wherein the second trunk has an associated head pointer and an associated tail pointer, and wherein sequentially writing the second PD metadata and/or data to the second trunk includes writing the second PD metadata and/or data to a head of the second trunk.
4. The method of claim 1 , further comprising:
in response to the first trunk having reached its storage capacity, allocating a third trunk in the shared PD tier of the logger; and
sequentially writing, by the primary storage node, third PD metadata and/or data to the third trunk.
5. The method of claim 4 , further comprising:
in response to the second trunk having reached its storage capacity, allocating a fourth trunk in the shared PD tier of the logger; and
sequentially writing, by the secondary storage node, fourth PD metadata and/or data to the fourth trunk.
6. The method of claim 1 , further comprising:
pushing the first PD metadata and/or data sequentially written to the first trunk to a first IO queue,
wherein merging the first PD metadata and/or data includes merging the first PD metadata and/or data in the first IO queue to obtain the first merged PD metadata and/or data.
7. The method of claim 6 , further comprising:
pushing the second PD metadata and/or data sequentially written to the second trunk to a second IO queue,
wherein merging the second PD metadata and/or data further includes merging the second PD metadata and/or data in the second IO queue to obtain the second merged PD metadata and/or data.
8. The method of claim 7 , wherein storing the first merged PD metadata and/or data and the second merged PD metadata and/or data includes storing the first merged PD metadata and/or data from the first IO queue to the NVM device, and storing the second merged PD metadata and/or data from the second IO queue to the NVM device.
9. The method of claim 1 , further comprising:
obtaining, by the secondary storage node, authorization from the primary storage node to write the second PD metadata and/or data to the shared PD tier of the logger.
10. The method of claim 9 , wherein allocating the second trunk in the shared PD tier of the logger includes allocating the second trunk in the shared PD tier in response to the authorization from the primary storage node being obtained.
11. An active-active clustered system, comprising:
a primary storage node;
a secondary storage node different from the primary storage node;
a first IO queue;
a second IO queue;
a logger including a shared page descriptor (PD) tier; and
a non-volatile memory (NVM) device,
wherein the primary storage node is configured to allocate at least a first trunk and a second trunk in the shared PD tier of the logger, and to sequentially write first PD metadata and/or data to the first trunk, wherein the first trunk and the second trunk are different trunks,
wherein the secondary storage node is configured to sequentially write second PD metadata and/or data to the second trunk,
wherein the logger is configured to be controlled for pushing the first PD metadata and/or data from the first trunk to the first IO queue, and for pushing the second PD metadata and/or data from the second trunk to the second IO queue,
wherein the first IO queue is configured to be controlled for merging the first PD metadata and/or data pushed from the first trunk to obtain first merged PD metadata and/or data, and for storing the first merged PD metadata and/or data to the NVM device, and
wherein the second IO queue is configured to be controlled for merging the second PD metadata and/or data pushed from the second trunk to obtain second merged PD metadata and/or data, and for storing the second merged PD metadata and/or data to the NVM device.
12. The system of claim 11 , wherein the first trunk has an associated head pointer and an associated tail pointer, and wherein the primary storage node is further configured to write the first PD metadata and/or data to a head of the first trunk.
13. The system of claim 12 , wherein the second trunk has an associated head pointer and an associated tail pointer, and wherein the secondary storage node is further configured to write the second PD metadata and/or data to a head of the second trunk.
14. The system of claim 11 , wherein the primary storage node is further configured, in response to the first trunk having reached its storage capacity, to allocate a third trunk in the shared PD tier of the logger, and to sequentially write third PD metadata and/or data to the third trunk.
15. The system of claim 14 , wherein the primary storage node is further configured, in response to the second trunk having reached its storage capacity, to allocate a fourth trunk in the shared PD tier of the logger.
16. The system of claim 15 , wherein the secondary storage node is further configured to sequentially write fourth PD metadata and/or data to the fourth trunk.
17. A computer program product including a set of non-transitory, computer-readable media having instructions that, when executed by storage processing circuitry, cause the storage processing circuitry to perform a method comprising:
allocating at least a first trunk and a second trunk in a shared page descriptor (PD) tier of a logger, wherein the first trunk and the second trunk are different trunks;
sequentially writing first PD metadata and/or data by a primary storage node to the first trunk, and sequentially writing second PD metadata and/or data by a secondary storage node to the second trunk;
merging the first PD metadata and/or data sequentially written to the first trunk to obtain first merged PD metadata and/or data, and merging the second PD metadata and/or data sequentially written to the second trunk to obtain second merged PD metadata and/or data; and
storing the first merged PD metadata and/or data and the second merged PD metadata and/or data to a non-volatile memory (NVM) device.
18. The computer program product of claim 17 , wherein the method further comprises:
pushing the first PD metadata and/or data sequentially written to the first trunk to a first IO queue,
wherein merging the first PD metadata and/or data includes merging the first PD metadata and/or data in the first IO queue to obtain the first merged PD metadata and/or data.
19. The computer program product of claim 18 , wherein the method further comprises:
pushing the second PD metadata and/or data sequentially written to the second trunk to a second IO queue,
wherein merging the second PD metadata and/or data further includes merging the second PD metadata and/or data in the second IO queue to obtain the second merged PD metadata and/or data.
20. The computer program product of claim 19 , wherein storing the first merged PD metadata and/or data and the second merged PD metadata and/or data includes storing the first merged PD metadata and/or data from the first IO queue to the NVM device, and storing the second merged PD metadata and/or data from the second IO queue to the NVM device.