[
https://issues.apache.org/jira/browse/HDDS-16602?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chung-En Lee updated HDDS-16602:
--------------------------------
Description:
This issue tracks the investigation and optimization of client request calls
via S3G.
We aim to:
* Investigate S3 API request paths (via OpenTelemetry traces) to identify
redundant or unnecessary RPC roundtrips.
* Eliminate these redundant RPCs across endpoints to reduce request latency
and OM load.
*Scope & Strategy*
Following the approach from HDDS-7324, we optimize endpoint execution by
directly assuming the S3 volume context, eliminating the redundant round-trip
to fetch it:
* Object-level operations: All object-level endpoints adopt this approach, as
the massive request volume at the object layer makes RPC elimination critical
for OM throughput.
* Bucket-level operations: We exclusively optimize \{{GET_BUCKET}}
(ListObjects), as it serves as the primary core bucket operation, intentionally
leaving the remaining bucket-level operations out of scope.
was:
This issue tracks the investigation and optimization of client requests calls
via S3g.
We aim to:
# Investigation S3 API request paths (by OpenTelemetry traces) to identify
redundant or unnecessary RPC roundtrips.
# Remove these redundant RPCs across endpoints to reduce request latency and
OM load.
*Discussion*
Consider extending the approach from HDDS-7324 by assuming the S3 volume
context, thereby avoiding the extra round-trip required by each S3 API to fetch
it.
> Reduce redundant Ozone RPC calls via s3g
> ----------------------------------------
>
> Key: HDDS-16602
> URL: https://issues.apache.org/jira/browse/HDDS-16602
> Project: Apache Ozone
> Issue Type: Improvement
> Reporter: Chung-En Lee
> Assignee: Chung-En Lee
> Priority: Major
>
> This issue tracks the investigation and optimization of client request calls
> via S3G.
> We aim to:
> * Investigate S3 API request paths (via OpenTelemetry traces) to identify
> redundant or unnecessary RPC roundtrips.
> * Eliminate these redundant RPCs across endpoints to reduce request latency
> and OM load.
> *Scope & Strategy*
> Following the approach from HDDS-7324, we optimize endpoint execution by
> directly assuming the S3 volume context, eliminating the redundant round-trip
> to fetch it:
> * Object-level operations: All object-level endpoints adopt this approach,
> as the massive request volume at the object layer makes RPC elimination
> critical for OM throughput.
> * Bucket-level operations: We exclusively optimize \{{GET_BUCKET}}
> (ListObjects), as it serves as the primary core bucket operation,
> intentionally leaving the remaining bucket-level operations out of scope.
>
>
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
