[Enhancement] "Wait and Fetch Result Time" in Point Lookup Queries with Multiple Fields and High Concurrency

[449300893]

Search before asking

• I had searched in the issues and found no similar issues.

Description

Execution Summary:

• Parse SQL Time: 38ms
• Nereids Lock Table Time: 40ms
• Nereids Analysis Time: 3ms
• Nereids Rewrite Time: 57ms
• Nereids Optimize Time: 22ms
• Nereids Translate Time: 1ms
• Workload Group: normal
• Analysis Time: 3ms
• Plan Time: 90ms
  • JoinReorder Time: N/A
  • CreateSingleNode Time: N/A
  • QueryDistributed Time: N/A
  • Init Scan Node Time: N/A
  • Finalize Scan Node Time: N/A
    • Get Splits Time: N/A
      • Get Partitions Time: N/A
      • Get Partition Files Time: N/A
    • Create Scan Range Time: N/A
• Schedule Time: 125ms
  • Fragment Assign Time: 1ms
  • Fragment Serialize Time: 57ms
  • Fragment RPC Phase1 Time: 59ms
  • Fragment RPC Phase2 Time: 8ms
  • Fragment Compressed Size: 180.20 KB
  • Fragment RPC Count: 6
• Schedule Time Of BE: {"phase1":{"8060":{"RPC Work Time":"41ms","RPC Latency From FE To BE":"0ms","RPC Work Queue Time":"0ms","RPC Latency From BE To FE":"17ms"},"8060":{"RPC Work Time":"36ms","RPC Latency From FE To BE":"0ms","RPC Work Queue Time":"0ms","RPC Latency From BE To FE":"22ms"},"8060":{"RPC Work Time":"50ms","RPC Latency From FE To BE":"1ms","RPC Work Queue Time":"0ms","RPC Latency From BE To FE":"8ms"}},"phase2":{"8060":{"RPC Work Time":"0ms","RPC Latency From FE To BE":"1ms","RPC Work Queue Time":"0ms","RPC Latency From BE To FE":"7ms"},"8060":{"RPC Work Time":"0ms","RPC Latency From FE To BE":"1ms","RPC Work Queue Time":"0ms","RPC Latency From BE To FE":"7ms"},"8060":{"RPC Work Time":"0ms","RPC Latency From FE To BE":"7ms","RPC Work Queue Time":"0ms","RPC Latency From BE To FE":"1ms"}}}
• Wait and Fetch Result Time: 4s837ms
  • Fetch Result Time: 4s833ms
  • Write Result Time: 0ms
• Doris Version: doris-2.1.8-1-834d802457
• Is Nereids: Yes
• Is Pipeline: Yes
• Is Cached: No
• Total Instances Num: 25
• Instances Num Per BE:
• Parallel Fragment Exec Instance Num: 4
• Trace ID:
• Transaction Commit Time: N/A
• Executed By Frontend: N/A
• Nereids GarbageCollect Time: 0ms
• Nereids BeFoldConst Time: 0ms
  查询场景：100并发，select 30+字段每次查询where in 1000条数据。请问有没有参数可以优化，感谢

Solution

No response

Are you willing to submit PR?

• Yes I am willing to submit a PR!

Code of Conduct

• I agree to follow this project's Code of Conduct

[ulekarsiddhant0-boop]

Hello @449300893 , I am interested in this issue #59534. I properly understood this issue. So, please assign this issue to me for working on it and create pr.
Thank You!

[449300893]

Hello @449300893 , I am interested in this issue #59534. I properly understood this issue. So, please assign this issue to me for working on it and create pr. Thank You!

hi ，Thanks for your interest in issue #59534!

Update: I tried to assign this issue to you, but unfortunately, my current permissions in the Apache Doris repository don't allow me to assign issues through the GitHub UI.

[449300893]

Could someone please take a look at this? Right now, the Wait and Fetch Result Time: 4s837ms is too long, which is causing the CPU usage to remain very low, and the efficiency is not improving.

[449300893]

The bottleneck occurs on the client side, specifically with the PreparedStatement parameter