Fast small write forwarding with non-temporal cache memory
A portion of the shared global memory of a storage array is allocated for write-only blocks. Writes to a same-block of a production device may be accumulated in the allocated portion of memory. Temporal sequencing may be associated with each accumulated version of the same-block. When idle processing resources become available, the oldest group of same-blocks may be consolidated based on the temporal sequencing. The consolidated block may then be destaged to cache slots or managed drives. A group of same-blocks may also be consolidated in response to a read command.
1. An apparatus comprising:
a storage array comprising:
a plurality of computing nodes that present a production device and manage access to persistent storage drives, each of the computing nodes comprising memory, a portion of the memory being allocated to a shared memory that can be accessed by other ones of the plurality of computing nodes, a first portion of the shared memory being allocated for write-only blocks and a second portion of the shared memory being allocated for readable and writable tracks, wherein the tracks are larger than the blocks; and
a block write service processor that writes a plurality of versions of a same-block of the production device to the first portion of the shared memory, each version having a temporal sequence associated therewith.
2. The apparatus of claim 1 wherein the block write service processor allocates a fixed-size buffer in the first portion of the shared memory in response to a write command from a host computer.
3. The apparatus of claim 2 wherein the block write service processor sends a transfer ready message to the host computer after the buffer is allocated.
4. The apparatus of claim 3 wherein the block write service processor receives a version of the same-block of data in response to the transfer ready message, and sends an ACK to the host computer after writing the version of the same-block into the allocated buffer.
5. The apparatus of claim 4 further comprising a block write aggregation processor that associates the temporal sequence with the corresponding version of the same-block.
6. The apparatus of claim 5 further comprising an aggregation scheduler, responsive to system idle feedback, to prompt the block write aggregation processor to consolidate a selected accumulated group of the plurality of versions of the same-block in the first portion of the shared memory.
7. The apparatus of claim 6 wherein the aggregation scheduler selects an oldest group of the plurality of versions of the same-block as the selected accumulated group.
8. The apparatus of claim 7 wherein the block write aggregation processor consolidates the selected accumulated group of the plurality of versions of the same-block based on the temporal sequence, thereby generating a consolidated block.
9. The apparatus of claim 8 wherein the block write aggregation processor destages the consolidated block to the persistent storage drives or the second portion of the shared memory.
10. The apparatus of claim 1 wherein, responsive to a command from a host computer to read the same-block that is present in the first portion of the shared memory, the block write aggregation processor consolidates the plurality of versions of the same-block based on the temporal sequence to generate a consolidated block, and returns the consolidated block to the host computer.
11. A method comprising:
in a storage array comprising a plurality of computing nodes that present a production device and manage access to persistent storage drives, each of the computing nodes comprising memory, a portion of the memory being allocated to a shared memory that can be accessed by other ones of the plurality of computing nodes:
allocating a first portion of the shared memory for write-only blocks;
allocating a second portion of the shared memory for readable and writable tracks, wherein the tracks are larger than the blocks;
writing a plurality of versions of a same-block of the production device to the first portion of the shared memory; and
associating a temporal sequence with each of the plurality of versions of the same-block.
12. The method of claim 11 comprising allocating a fixed-size buffer in the first portion of the shared memory in response to a write command from a host computer.
13. The method of claim 12 comprising sending a transfer ready message to the host computer after the buffer is allocated.
14. The method of claim 13 comprising receiving a version of the same-block of data in response to the transfer ready message, and sending an ACK to the host computer after writing the version of the same-block into the allocated buffer.
15. The method of claim 14 comprising a block write aggregation processor associating the temporal sequence with the corresponding version of the same-block.
16. The method of claim 15 comprising, responsive to system idle feedback, consolidating a selected accumulated group of the plurality of versions of the same-block in the first portion of the shared memory.
17. The method of claim 16 comprising selecting an oldest group of the plurality of versions of the same-block as the selected accumulated group.
18. The method of claim 17 comprising consolidating the selected accumulated group of the plurality of versions of the same-block based on the temporal sequence, thereby generating a consolidated block.
19. The method of claim 18 comprising destaging the consolidated block to the persistent storage drives or the second portion of the shared memory.
20. The method of claim 11 comprising, responsive to a command from a host computer to read the same-block that is present in the first portion of the shared memory, consolidating the plurality of versions of the same-block based on the temporal sequence to generate a consolidated block, and returning the consolidated block to the host computer.