[feature](shuffle) enable strict consistency dml by default #32958
Conversation
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
|
Thank you for your contribution to Apache Doris. Since 2024-03-18, the Document has been moved to doris-website. |
|
run buildall |
TPC-H: Total hot run time: 37802 ms |
TPC-DS: Total hot run time: 182340 ms |
|
Load test result on machine: 'aliyun_ecs.c7a.8xlarge_32C64G' |
|
run buildall |
|
run buildall |
1 similar comment
|
run buildall |
TPC-H: Total hot run time: 38533 ms |
morrySnow
previously approved these changes
Apr 23, 2024
github-actions
bot
added
the
approved
|
PR approved by at least one committer and no changes requested. |
|
PR approved by anyone and no changes requested. |
|
run buildall |
github-actions
bot
removed
the
approved
|
run buildall |
5 similar comments
|
run buildall |
|
run buildall |
|
run buildall |
|
run buildall |
|
run buildall |
TPC-H: Total hot run time: 41764 ms |
TPC-DS: Total hot run time: 186417 ms |
morningman
reviewed
Apr 29, 2024
| @@ -636,6 +636,12 @@ public class Config extends ConfigBase { | |||
| varType = VariableAnnotation.EXPERIMENTAL) | |||
| public static boolean enable_single_replica_load = false; | |||
|
|
|||
| @ConfField(mutable = true, masterOnly = true, description = { | |||
| "对于 DUPLICATE KEY 表启用 shuffle 的最小 tablet 数量", | |||
There was a problem hiding this comment.
Could you explain more about when should user set this config?
There was a problem hiding this comment.
Enable shuffle can bring both positive effect and negative effect.
Shuffle must be enabled for UNIQ and AGG tables for data consistency reasons.
But for DUP tables, we can do some trade-offs.
If shuffle is disabled, the load will be faster but it will use more memory.
If shuffle is enabled, the load will be slower but it will use less memory.
When loading into table with more buckets, it's more likely to be memory constrained.
So we choose to enable shuffle only for tables with certain amount buckets.
TPC-DS: Total hot run time: 186949 ms |
kaijchen
commented
May 6, 2024
kaijchen
commented
May 6, 2024
|
run buildall |
TPC-H: Total hot run time: 40818 ms |
TPC-DS: Total hot run time: 185397 ms |
|
run buildall |
TPC-DS: Total hot run time: 186939 ms |
|
run buildall |
github-actions
bot
added
the
approved
|
PR approved by at least one committer and no changes requested. |
|
run buildall |
yiguolei
pushed a commit
that referenced
this pull request
May 8, 2024
kaijchen
added a commit
to kaijchen/doris
that referenced
this pull request
May 10, 2024
yiguolei
pushed a commit
that referenced
this pull request
May 10, 2024
ByteYue
pushed a commit
to ByteYue/doris
that referenced
this pull request
May 15, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Proposed changes
Set
enable_strict_consistenct_dmltotrueby default.Enable shuffle can bring both positive effect and negative effect.
Shuffle must be enabled for
UNIQandAGGtables for data consistency reasons.But for
DUPtables, we can do some trade-offs.If shuffle is disabled, the load will be faster but it will use more memory.
If shuffle is enabled, the load will be slower but it will use less memory.
When loading into table with more buckets, it's more likely to be memory constrained.
So we choose to enable shuffle only for tables with certain amount buckets.
The threshold is set by
min_tablets_for_dup_table_shufflein be.conf.Further comments
If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...