This document describes the performance test results of a general TencentDB for MySQL 5.6 instance.
In the full cache scenario, as all data can be put into the cache, the disk doesn't need to be read and written to update the cache during queries.
CPU (Core) |
Memory (MB) |
Concurrency | Data Volume/Table | Total Tables | SysBench TPS | SysBench QPS | avg_lat |
---|---|---|---|---|---|---|---|
1 | 1000 | 8 | 25000 | 150 | 577.97 | 11559.4 | 13.84 |
1 | 2000 | 8 | 25000 | 150 | 550.21 | 11004.3 | 14.54 |
2 | 4000 | 16 | 25000 | 150 | 1190.41 | 23808.1 | 13.44 |
4 | 8000 | 32 | 25000 | 150 | 2307.1 | 46142 | 13.87 |
4 | 16000 | 32 | 25000 | 150 | 2229.8 | 44596.1 | 14.35 |
8 | 16000 | 64 | 25000 | 150 | 3771.25 | 75425 | 16.97 |
8 | 32000 | 64 | 25000 | 150 | 3909.36 | 78187.1 | 16.37 |
16 | 32000 | 128 | 25000 | 150 | 6102.35 | 122047 | 20.97 |
16 | 64000 | 128 | 25000 | 150 | 6788.83 | 135777 | 18.85 |
16 | 96000 | 128 | 25000 | 150 | 6771.81 | 135436 | 18.9 |
16 | 128000 | 128 | 25000 | 150 | 7039.65 | 140793 | 18.18 |
In the disk I/O scenario, as only part of the data can be put into the cache, the disk needs to be read and written to update the cache during queries.
CPU (Core) |
Memory (MB) |
Concurrency | Data Volume/Table | Total Tables | SysBench TPS | SysBench QPS | avg_lat |
---|---|---|---|---|---|---|---|
1 | 1000 | 8 | 800000 | 6 | 530.36 | 10607.3 | 15.08 |
1 | 2000 | 8 | 800000 | 12 | 543.86 | 10877.2 | 14.71 |
2 | 4000 | 16 | 800000 | 24 | 1120.47 | 22409.4 | 14.28 |
4 | 8000 | 32 | 800000 | 48 | 2189.87 | 43797.4 | 14.61 |
4 | 16000 | 32 | 6000000 | 13 | 2138.75 | 42775.1 | 14.96 |
8 | 16000 | 64 | 6000000 | 13 | 3589.44 | 71788.8 | 17.83 |
8 | 32000 | 64 | 6000000 | 25 | 3535.54 | 70710.8 | 18.1 |
16 | 32000 | 128 | 6000000 | 25 | 5550.31 | 111006 | 23.06 |
16 | 64000 | 128 | 6000000 | 49 | 6414.39 | 128288 | 19.95 |
16 | 96000 | 128 | 6000000 | 74 | 5874.64 | 117493 | 21.78 |
16 | 128000 | 128 | 6000000 | 98 | 5611.06 | 112221 | 22.81 |
Was this page helpful?