You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@drill.apache.org by "Bowen Ding (Jira)" <ji...@apache.org> on 2020/06/26 13:44:00 UTC
[jira] [Created] (DRILL-7753) Storage IPFS: Better peer selection
algorithm
Bowen Ding created DRILL-7753:
---------------------------------
Summary: Storage IPFS: Better peer selection algorithm
Key: DRILL-7753
URL: https://issues.apache.org/jira/browse/DRILL-7753
Project: Apache Drill
Issue Type: Improvement
Components: Storage - Other
Reporter: Bowen Ding
Assignee: Bowen Ding
The current implementation randomly selects one provider and assigns it as the executor responsible for executing the query. It does not check reachability, latency, bandwidth between the foreman and the remote peer, etc.
A better peer selection algorithm ideally takes into account the above factors, in order to minimize the total query execution time.
As a subset of the bigger issue, if a provider has multiple advertised `Multiaddress`es, it is also important to know what network address to use to do networking. The current implementation prefers globally routable addresses over local-only addresses, and does not check reachability.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)