Skip to content

Conversation

@before-Sunrise
Copy link

No description provided.

Signed-off-by: before-Sunrise <unclejyj@gmail.com>
@CLAassistant
Copy link

CLAassistant commented Dec 25, 2025

CLA assistant check
All committers have signed the CLA.

Signed-off-by: before-Sunrise <unclejyj@gmail.com>
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are already result json files for six system. As per this PR,

  • a new configuration is added which is basically very similar to existing configuration c7a.metal-48xl.json
  • the existing six configurations are not updated.

Can I recommend to remove c8g.metal-48xl.json and instead update the existing ones? (or let us do the latter)?

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There are already result json files for six system. As per this PR,

  • a new configuration is added which is basically very similar to existing configuration c7a.metal-48xl.json
  • the existing six configurations are not updated.

Can I recommend to remove c8g.metal-48xl.json and instead update the existing ones? (or let us do the latter)?

so if i want to add a new configuration for a new version of starrocks, I need to update every exsited configurations? I want to keep the c8g.metal-48xl's result, so can you help to update the other configuration based on starrocks 4.0.2?

)
DUPLICATE KEY (CounterID, EventDate, UserID, EventTime, WatchID)
DISTRIBUTED BY RANDOM BUCKETS 1
DISTRIBUTED BY HASH(UserID) BUCKETS 192
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For my understanding: Why does the partitioning scheme change? ClickBench runs on a single node, so hash partitioning probably doesn't make a difference at all.

Copy link
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For my understanding: Why does the partitioning scheme change? ClickBench runs on a single node, so hash partitioning probably doesn't make a difference at all.

This is the most common table creation method for users in starrocks community. First, hot and cold partitioning is based on time, and then hash bucketing is performed based on users. I think this can better reflect the starrocks' performance. I also noticed the apache doris use hash bucketing for table creation and clickbench accepted it, so I think starrocks can use hash bucketing too

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants