Summary:
For our default block cache, each additional entry has extra memory overhead. It include LRUHandle (72 bytes currently) and the cache key (two varint64, file id and offset). The usage is not negligible. For example for block_size=4k, the overhead accounts for an extra 2% memory usage for the cache. The patch charging the cache for the extra usage, reducing untracked memory usage outside block cache. The feature is enabled by default and can be disabled by passing kDontChargeCacheMetadata to the cache constructor.
This PR builds up on https://github.com/facebook/rocksdb/issues/4258
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5797
Test Plan:
- Existing tests are updated to either disable the feature when the test has too much dependency on the old way of accounting the usage or increasing the cache capacity to account for the additional charge of metadata.
- The Usage tests in cache_test.cc are augmented to test the cache usage under kFullChargeCacheMetadata.
Differential Revision: D17396833
Pulled By: maysamyabandeh
fbshipit-source-id: 7684ccb9f8a40ca595e4f5efcdb03623afea0c6f
Summary:
This PR adds four more eviction policies.
- OPT [1]
- Hyperbolic caching [2]
- ARC [3]
- GreedyDualSize [4]
[1] L. A. Belady. 1966. A Study of Replacement Algorithms for a Virtual-storage Computer. IBM Syst. J. 5, 2 (June 1966), 78-101. DOI=http://dx.doi.org/10.1147/sj.52.0078
[2] Aaron Blankstein, Siddhartha Sen, and Michael J. Freedman. 2017. Hyperbolic caching: flexible caching for web applications. In Proceedings of the 2017 USENIX Conference on Usenix Annual Technical Conference (USENIX ATC '17). USENIX Association, Berkeley, CA, USA, 499-511.
[3] Nimrod Megiddo and Dharmendra S. Modha. 2003. ARC: A Self-Tuning, Low Overhead Replacement Cache. In Proceedings of the 2nd USENIX Conference on File and Storage Technologies (FAST '03). USENIX Association, Berkeley, CA, USA, 115-130.
[4] N. Young. The k-server dual and loose competitiveness for paging. Algorithmica, June 1994, vol. 11,(no.6):525-41. Rewritten version of ''On-line caching as cache size varies'', in The 2nd Annual ACM-SIAM Symposium on Discrete Algorithms, 241-250, 1991.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5644
Differential Revision: D16548817
Pulled By: HaoyuHuang
fbshipit-source-id: 838f76db9179f07911abaab46c97e1c929cfcd63
Summary:
This PR optimizes the hybrid row-block cache simulator. If a Get request hits the cache, we treat all its future accesses as hits.
Consider a Get request (no snapshot) accesses multiple files, e.g, file1, file2, file3. We construct the row key as "fdnumber_key_0". Before this PR, if it hits the cache when searching the key in file1, we continue to process its accesses in file2 and file3 which is unnecessary.
With this PR, if "file1_key_0" is in the cache, we treat all future accesses of this Get request as hits.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5616
Differential Revision: D16453187
Pulled By: HaoyuHuang
fbshipit-source-id: 56f3169cc322322305baaf5543226a0824fae19f
Summary:
- Compute correlation between a few features and predictions, e.g., number of accesses since the last access vs number of accesses till the next access on a block.
- Output human readable trace file so python can consume it.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5596
Test Plan: make clean && USE_CLANG=1 make check -j32
Differential Revision: D16373200
Pulled By: HaoyuHuang
fbshipit-source-id: c848d26bc2e9210461f317d7dbee42d55be5a0cc
Summary:
This PR traces the referenced key for Get for all types of blocks. This is useful when evaluating hybrid row-block caches.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5548
Test Plan: make clean && USE_CLANG=1 make check -j32
Differential Revision: D16157979
Pulled By: HaoyuHuang
fbshipit-source-id: f6327411c9deb74e35e22a35f66cdbae09ab9d87
Summary:
This PR adds a ghost cache for admission control. Specifically, it admits an entry on its second access.
It also adds a hybrid row-block cache that caches the referenced key-value pairs of a Get/MultiGet request instead of its blocks.
Pull Request resolved: https://github.com/facebook/rocksdb/pull/5534
Test Plan: make clean && COMPILE_WITH_ASAN=1 make check -j32
Differential Revision: D16101124
Pulled By: HaoyuHuang
fbshipit-source-id: b99edda6418a888e94eb40f71ece45d375e234b1
Maysam Yabandeh