xianjingfeng opened a new issue, #341: URL: https://github.com/apache/incubator-uniffle/issues/341
### Code of Conduct - [X] I agree to follow this project's [Code of Conduct](https://www.apache.org/foundation/policies/conduct) ### Search before asking - [X] I have searched in the [issues](https://github.com/apache/incubator-uniffle/issues?q=is%3Aissue) and found no similar issues. ### Describe the feature Support rack awareness for better performence. ### Motivation If our tasks and shuffle servers are not in same rack or same data center,the load of the network switch will be very high, and the performance will be greatly affected. So we need support that the tasks send data to the closest shuffle servers as much as possible. ### Describe the solution 1. Assign rack information to each shuffle server and each compute nodes(such as Yarn Nodemanager/Kubernetes Node). We can store this information to hdfs or other external storage and manage by coordinator. 2. Implement a new `AssignmentStrategy` that assign nodes in different racks as much as possible. 3. Client get its own rack information from coordinator and send data to the closest shuffle servers as much as possible. ### Additional context _No response_ ### Are you willing to submit PR? - [X] Yes I am willing to submit a PR! -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
