MongoDB 3.2 Performance Test

Test Environment machine configuration

Linux container

• 4c/16g/300gssd
• 8c/32g/300gssd

Test object

version	engine	parameter Configuration
		4c/16g	8c/32g
mongodb3.2.6	Wiredtiger	Cachesizegb:12 Syncperiodsecs:1 Collectionconfig: Blockcompressor:snappy Indexconfig: Prefixcompression:true	Cachesizegb:24 Syncperiodsecs:1 Collectionconfig: Blockcompressor:snappy Indexconfig: Prefixcompression:true
tokumx1.5	Tokumx	cachesize=12g Syncdelay=5	cachesize=24g Syncdelay=5
tokumx2.0.2	Tokumx	cachesize=12g checkpointperiod=10 cleaneriterations=10 Directio=false cleanerperiod=2 Syncdelay=5	cachesize=24g checkpointperiod=10 cleaneriterations=10 Directio=false cleanerperiod=2 Syncdelay=5

Test scenario

• Test a single-node environment

1. 100%insert
2. Single-node _50%update50%read
3. 5%update5%insert90%read
4. 100%read
5. Comparison of wiredtiger_syncperiodsecs_60 with 1
6. Sharding Cluster Performance Pressure measurement

• Description

1. Scenario 1-4 1000W data per load, data size approx. 13G
2. Scenario 5 load 5000W data, data size approx. 75G

Test method

• YCSB Pressure Measurement

Test Result Scenario 1: Single node _100%insert (load data) Scenario 2: Single node _50%update50%read

Scenario 3: Single-node _5%update5%insert90%read Scenario 4: Single-node _100%read scene 5:wiredtiger_syncperiodsecs_60_1

Scenario 6:sharding cluster Performance test

Conclusion

• Load performance comparison, Wiredtiger advantage is very obvious, the speed is approximately the same as the configuration of the Tokumx 5 times times, and RT Shorter
• Read-only performance, Wiredtiger performance and tokumx, comparison, poor performance, but stable;
• In complex cases, Wiredtiger has better performance and can support the thread invocation of higher concurrency;
• Without disk-based and network throughput considerations, the sharding of three of the following nodes has no performance value, and at this stage it appears that deploying MONGOs as much as possible can effectively improve overall cluster utilization.

MongoDB 3.2 Performance Test