Techniques for efficiently organizing storage of compressed extents

1. A method of efficiently storing compressed data of a storage object in a data storage system, the method comprising:

receiving, in a cache buffer, a number, U, of uncompressed data blocks of a uniform size, the uncompressed data blocks received within write requests directed to the storage object;

compressing the U uncompressed data blocks of the cache buffer into respective compressed extents;

performing an optimization operation including generating a set of distributions of compressed extents among a plurality of containers and searching the set for a distribution having a minimal total amount of storage taken up by its respective plurality of containers, each container having a respective size equal to a respective integer multiple of the uniform size of the uncompressed data blocks; and

storing the compressed extents within a plurality of containers in persistent storage in accordance with the distribution having the minimal total amount of storage taken up by its respective plurality of containers.

2. The method of claim 1 wherein each container is configured to store no more than a predefined maximum allowable number of compressed extents.

3. The method of claim 2 wherein generating the set includes excluding from consideration any compressed extent that is not at least 512 bytes smaller than the uncompressed data block from which it was compressed.

4. The method of claim 2 wherein compressing the U uncompressed data blocks of the cache buffer into respective compressed extents includes prefixing each compressed extent with a fixed-size header that describes that compressed extent.

5. The method of claim 4 wherein performing the optimization operation includes determining a compressed size of each compressed extent including the prefixed header and rounding up to the next integer multiple of 512 bytes.

6. The method of claim 5 wherein performing the optimization operation further includes:

sorting the compressed extents in order of their compressed sizes; and

assigning the compressed extents into the plurality of segments using one of a first fit decreasing heuristic, a best fit decreasing heuristic, and a modified first fit decreasing heuristic.

7. The method of claim 5 wherein performing the optimization operation further includes:

summing the compressed size of each compressed extent to yield a sum;

rounding the sum up to the next integer multiple of the uniform size of the uncompressed data blocks to yield a starting size; and

generating and searching a space of distributions among pluralities of containers having a combined size equal to the starting size and iteratively increasing the combined size of the space until a solution is found.

8. The method of claim 2 wherein performing the optimization operation includes searching the set of distributions using a greedy technique.

9. The method of claim 2 wherein receiving the U uncompressed data blocks includes receiving at least 3 times the predefined maximum and no more than 4 times the predefined maximum.

10. The method of claim 2 wherein the predefined maximum is less than 100.

11. The method of claim 2 wherein the predefined maximum is equal to 12 and U is equal to 36.

12. The method of claim 1 wherein storing the compressed extents within the plurality of containers in the persistent storage includes storing, for each of the plurality of containers, a respective mapping structure within the persistent storage, the mapping structure mapping particular blocks of the storage object to particular offsets within that container.

13. An apparatus for efficiently storing compressed data of a storage object, the apparatus comprising:

persistent storage;

network interface circuitry for connecting to a network; and

processing circuitry coupled to memory configured to:

receive, from the network interface circuitry, in a cache buffer of the memory, a number, U, of uncompressed data blocks of a uniform size, the uncompressed data blocks received within write requests directed to the storage object;

compress the U uncompressed data blocks of the cache buffer into respective compressed extents;

performing an optimization operation including generating a set of distributions of compressed extents among a plurality of containers and searching the set for a distribution having a minimal total amount of storage taken up by its respective plurality of containers, each container having a respective size equal to a respective integer multiple of the uniform size of the uncompressed data blocks; and

storing the compressed extents within a plurality of containers in the persistent storage in accordance with the distribution having the minimal total amount of storage taken up by its respective plurality of containers.

14. A computer program product comprising a non-transitory computer-readable storage medium storing a set of instructions, which, when executed by a computing device, causes the computing device to store compressed data of a storage object in persistent storage by:

receiving, in a cache buffer, a number, U, of uncompressed data blocks of a uniform size, the uncompressed data blocks received in write requests directed to the storage object;

compressing the U uncompressed data blocks of the cache buffer into respective compressed extents;

performing an optimization operation including generating a set of distributions of compressed extents among a plurality of containers and searching the set for a distribution having a minimal total amount of storage taken up by its respective plurality of containers, each container having a respective size equal to a respective integer multiple of the uniform size of the uncompressed data blocks; and

storing the compressed extents within a plurality of containers in the persistent storage in accordance with the distribution having the minimal total amount of storage taken up by its respective plurality of containers.