[SERVER-74268] Run YCSB + Config ops Created: 22/Feb/23  Updated: 07/Jun/23  Resolved: 07/Jun/23

Status: Closed
Project: Core Server
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Task Priority: Major - P3
Reporter: Kshitij Gupta Assignee: Wenqin Ye
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified

Assigned Teams:
Sharding NYC
Sprint: Sharding NYC 2023-05-01, Sharding NYC 2023-05-15, Sharding NYC 2023-05-29, Sharding NYC 2023-06-12
Participants:

 Description   

Config ops like DDL operations and chunk migrations.



 Comments   
Comment by Wenqin Ye [ 07/Jun/23 ]

We used the same YCSB workload as SERVER-74267 (56 million starting documents with 20 million operations with various read/write percentages).

To perform the config ops, we modified the balancer code to initiate random chunk migrations every 10 seconds. Also the config server’s hardware was upgraded to match that of a dedicated shard server.

The averaged results from three runs of the YCSB workload with config shard and config ops were compared to the averaged results from three runs of the YCSB workload without config shard but with config ops.

The results can be found in this spreadsheet: https://docs.google.com/spreadsheets/d/1tezs-_nY6l84arS0y_HGxF_gxtCs-h3yRDqUDP4lrsI/edit#gid=1773702825

Overall there were no statistically significant performance regressions. The performance data was very noisy due to the randomness of the chunk migrations; however, all performance regressions were within 1 standard deviation of 0% performance regressions (if there was a standard deviation). This suggests that the performance regressions observed were likely due to random noise and not performance differences between YCSB with config shard and without config shard.

Generated at Thu Feb 08 06:26:58 UTC 2024 using Jira 9.7.1#970001-sha1:2222b88b221c4928ef0de3161136cc90c8356a66.