[ https://issues.apache.org/jira/browse/SOLR-9636?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15630883#comment-15630883 ]
Erick Erickson commented on SOLR-9636: -------------------------------------- I've spent too much of my life fixing the wrong thing for efficiency, I've got to agree with Joel here. Let's positively identify the bottleneck before proposing a solution. SOLR-486 does show roughly a doubling of binary output compared to JSON, my uncertainty is how much of that improvement is actually forming the response. An interesting measurement would be to just avoid assembling _anything_ for the output stream, just read the date from the docValues and throw it away. But even there it's iffy. My testing purposely threw the bits on the floor, so network transmission was completely out of the picture. Whether the end user would see measurable improvement is an open question. Contrariwise, see https://issues.apache.org/jira/browse/SOLR-9296 where I was doing some experimenting with re-using things like char[] rather than allocating new strings all the time and saw up to an 11% improvement in throughput. Don't know whether the notion of going to a binary format would make that irrelevant that or not. I'll link the two issues anyway just so don't lose the association. > Use javabin for /stream, /sql internode communication > ----------------------------------------------------- > > Key: SOLR-9636 > URL: https://issues.apache.org/jira/browse/SOLR-9636 > Project: Solr > Issue Type: Improvement > Security Level: Public(Default Security Level. Issues are Public) > Reporter: Noble Paul > Assignee: Noble Paul > Attachments: SOLR-9636.patch > > > currently it uses json, which is verbose and slow -- This message was sent by Atlassian JIRA (v6.3.4#6332) --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org