Last active
August 29, 2015 14:06
-
-
Save hyc/7593d1bf21a804c1c5be to your computer and use it in GitHub Desktop.
Samba NTDB microbench
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| https://github.com/hyc/leveldb/commit/d05251bb51138f8f77a08e0e01f22c5048d3ccb5 | |
| violino:/home/software/leveldb> ./db_bench_ntdb | |
| NTDB: version 1.0 | |
| Date: Mon Sep 15 16:51:55 2014 | |
| CPU: 4 * Intel(R) Core(TM)2 Extreme CPU Q9300 @ 2.53GHz | |
| CPUCache: 6144 KB | |
| Keys: 16 bytes each | |
| Values: 100 bytes each (50 bytes after compression) | |
| Entries: 1000000 | |
| RawSize: 110.6 MB (estimated) | |
| FileSize: 62.9 MB (estimated) | |
| ------------------------------------------------ | |
| fillrandsync : 90.975 micros/op 10992 ops/sec; 1.2 MB/s (1000 ops) | |
| 272 /tmp/leveldbtest-1000/dbbench_ntdb-1 | |
| 272 /tmp/leveldbtest-1000 | |
| fillrandom : 458.085 micros/op 2182 ops/sec; 0.2 MB/s | |
| 98560 /tmp/leveldbtest-1000/dbbench_ntdb-2 | |
| 98560 /tmp/leveldbtest-1000 | |
| fillrandbatch : 411.020 micros/op 2432 ops/sec; 0.3 MB/s | |
| 106752 /tmp/leveldbtest-1000/dbbench_ntdb-3 | |
| 106752 /tmp/leveldbtest-1000 | |
| fillseqsync : 91.421 micros/op 10938 ops/sec; 1.2 MB/s (1000 ops) | |
| 272 /tmp/leveldbtest-1000/dbbench_ntdb-4 | |
| 272 /tmp/leveldbtest-1000 | |
| fillseq : 995.226 micros/op 1004 ops/sec; 0.1 MB/s | |
| 149072 /tmp/leveldbtest-1000/dbbench_ntdb-5 | |
| 149072 /tmp/leveldbtest-1000 | |
| fillseqbatch : 1158.379 micros/op 863 ops/sec; 0.1 MB/s | |
| 155040 /tmp/leveldbtest-1000/dbbench_ntdb-6 | |
| 155040 /tmp/leveldbtest-1000 | |
| overwrite : 116.464 micros/op 8586 ops/sec; 0.9 MB/s | |
| 155040 /tmp/leveldbtest-1000/dbbench_ntdb-6 | |
| 155040 /tmp/leveldbtest-1000 | |
| readrandom : 3.662 micros/op 273081 ops/sec; (1000000 of 1000000 found) | |
| readseq : 1.532 micros/op 652888 ops/sec; 72.2 MB/s | |
| violino:/home/software/leveldb> ./db_bench_mdb | |
| LMDB: version LMDB 0.9.14: (July 24, 2014) | |
| Date: Mon Sep 15 17:54:21 2014 | |
| CPU: 4 * Intel(R) Core(TM)2 Extreme CPU Q9300 @ 2.53GHz | |
| CPUCache: 6144 KB | |
| Keys: 16 bytes each | |
| Values: 100 bytes each (50 bytes after compression) | |
| Entries: 1000000 | |
| RawSize: 110.6 MB (estimated) | |
| FileSize: 62.9 MB (estimated) | |
| ------------------------------------------------ | |
| fillrandsync : 16.965 micros/op 58944 ops/sec; 6.5 MB/s (1000 ops) | |
| 228 /tmp/leveldbtest-1000/dbbench_mdb-1 | |
| 228 /tmp/leveldbtest-1000 | |
| fillrandom : 4.226 micros/op 236654 ops/sec; 26.2 MB/s | |
| 116580 /tmp/leveldbtest-1000/dbbench_mdb-2 | |
| 116580 /tmp/leveldbtest-1000 | |
| fillrandbatch : 3.248 micros/op 307882 ops/sec; 34.1 MB/s | |
| 126408 /tmp/leveldbtest-1000/dbbench_mdb-3 | |
| 126408 /tmp/leveldbtest-1000 | |
| fillseqsync : 12.043 micros/op 83035 ops/sec; 9.2 MB/s (1000 ops) | |
| 172 /tmp/leveldbtest-1000/dbbench_mdb-4 | |
| 172 /tmp/leveldbtest-1000 | |
| fillseq : 2.167 micros/op 461388 ops/sec; 51.0 MB/s | |
| 125872 /tmp/leveldbtest-1000/dbbench_mdb-5 | |
| 125872 /tmp/leveldbtest-1000 | |
| fillseqbatch : 0.462 micros/op 2166513 ops/sec; 239.7 MB/s | |
| 125872 /tmp/leveldbtest-1000/dbbench_mdb-6 | |
| 125872 /tmp/leveldbtest-1000 | |
| overwrite : 4.776 micros/op 209391 ops/sec; 23.2 MB/s | |
| 125872 /tmp/leveldbtest-1000/dbbench_mdb-6 | |
| 125872 /tmp/leveldbtest-1000 | |
| readrandom : 1.289 micros/op 775712 ops/sec; (1000000 of 1000000 found) | |
| readseq : 0.060 micros/op 16686412 ops/sec; 1846.0 MB/s | |
| readreverse : 0.054 micros/op 18510977 ops/sec; 2047.8 MB/s |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
LMDB benchmark seems promising.. pretty impressive!! LMDB has 1.36 million reads per second so all actual data resides in memory right or SSD? If yes do you have numbers read/write performance on SSD?What is the memory footprint of index (metadata) for the data and actual data and its comparison with levelDB / TDB? RocksDB is not in the comparison, is it because of poor performance?