gogrlx/bitcask
1
0
Fork 0
mirror of https://github.com/gogrlx/bitcask.git synced 2026-04-02 02:58:59 -07:00
Code Issues Packages Projects Releases Wiki Activity

Improved read/write performance by another ~2x by not calling Stat() on every read/write

Browse Source
This commit is contained in:
James Mills
2019-03-16 08:15:07 +10:00
parent 2585222830
commit c0f178c4f7
3 changed files with 54 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files

42
README.md
View File

@@ -97,33 +97,33 @@ Benchmarks run on a 11" Macbook with a 1.4Ghz Intel Core i7:
```
$ make bench
...
BenchmarkGet/128B-4 300000 5144 ns/op 400 B/op 5 allocs/op
BenchmarkGet/256B-4 300000 5166 ns/op 656 B/op 5 allocs/op
BenchmarkGet/512B-4 300000 5284 ns/op 1200 B/op 5 allocs/op
BenchmarkGet/1K-4 200000 5779 ns/op 2288 B/op 5 allocs/op
BenchmarkGet/2K-4 200000 6396 ns/op 4464 B/op 5 allocs/op
BenchmarkGet/4K-4 200000 7716 ns/op 9072 B/op 5 allocs/op
BenchmarkGet/8K-4 200000 9802 ns/op 17776 B/op 5 allocs/op
BenchmarkGet/16K-4 100000 13299 ns/op 34928 B/op 5 allocs/op
BenchmarkGet/32K-4 100000 21819 ns/op 73840 B/op 5 allocs/op
BenchmarkGet/128B-4 300000 5178 ns/op 400 B/op 5 allocs/op
BenchmarkGet/256B-4 300000 5273 ns/op 656 B/op 5 allocs/op
BenchmarkGet/512B-4 200000 5368 ns/op 1200 B/op 5 allocs/op
BenchmarkGet/1K-4 200000 5800 ns/op 2288 B/op 5 allocs/op
BenchmarkGet/2K-4 200000 6766 ns/op 4464 B/op 5 allocs/op
BenchmarkGet/4K-4 200000 7857 ns/op 9072 B/op 5 allocs/op
BenchmarkGet/8K-4 200000 9538 ns/op 17776 B/op 5 allocs/op
BenchmarkGet/16K-4 100000 13188 ns/op 34928 B/op 5 allocs/op
BenchmarkGet/32K-4 100000 21620 ns/op 73840 B/op 5 allocs/op
BenchmarkPut/128B-4 100000 12746 ns/op 825 B/op 8 allocs/op
BenchmarkPut/256B-4 100000 12937 ns/op 954 B/op 8 allocs/op
BenchmarkPut/512B-4 100000 14610 ns/op 1245 B/op 8 allocs/op
BenchmarkPut/1K-4 100000 16920 ns/op 1825 B/op 8 allocs/op
BenchmarkPut/2K-4 100000 22075 ns/op 2987 B/op 8 allocs/op
BenchmarkPut/4K-4 30000 40544 ns/op 5566 B/op 8 allocs/op
BenchmarkPut/8K-4 20000 63392 ns/op 10210 B/op 8 allocs/op
BenchmarkPut/16K-4 10000 108667 ns/op 19244 B/op 8 allocs/op
BenchmarkPut/32K-4 10000 129256 ns/op 41920 B/op 8 allocs/op
BenchmarkPut/128B-4 200000 7875 ns/op 409 B/op 6 allocs/op
BenchmarkPut/256B-4 200000 8712 ns/op 538 B/op 6 allocs/op
BenchmarkPut/512B-4 200000 9832 ns/op 829 B/op 6 allocs/op
BenchmarkPut/1K-4 100000 13105 ns/op 1410 B/op 6 allocs/op
BenchmarkPut/2K-4 100000 18601 ns/op 2572 B/op 6 allocs/op
BenchmarkPut/4K-4 50000 36631 ns/op 5151 B/op 6 allocs/op
BenchmarkPut/8K-4 30000 56128 ns/op 9798 B/op 6 allocs/op
BenchmarkPut/16K-4 20000 83209 ns/op 18834 B/op 6 allocs/op
BenchmarkPut/32K-4 10000 135899 ns/op 41517 B/op 6 allocs/op
BenchmarkScan-4 1000000 1858 ns/op 493 B/op 25 allocs/op
BenchmarkScan-4 1000000 1851 ns/op 493 B/op 25 allocs/op
```
For 128B values:
* ~180,000 reads/sec
* ~80,000 writes/sec
* ~200,000 reads/sec
* ~130,000 writes/sec
The full benchmark above shows linear performance as you increase key/value sizes.