time to return confluence search result

Ollie Guan

Hi Community,

we just upgraded to Confluence Data Center v7.13, our database is about 400GB, and the attachment is 1.5TB.

When we query in Confluence, the time to return results of different orders of magnitude

1. returns 4000 results, about 0.5 seconds;

2. returns 40000 results, about 1-2 seconds;

3. returns 400,000 results, about 4-6 seconds;

I want to know whether the system is possible to optimize?

Nic Brough -Adaptavist-

There's a couple of things you might look at here. 

The Confluence search mostly works from the index, and that is built for searching over delivering content, so one thing is that you can optimise searches by looking for things that are easier to index and search than others.  One simple example is searching for labels rather than just words - there's a lot more distinct words in the body text of your pages than there are distinct label words and phrases.  Another one is Confluence prioritises the age of pages (last update, not creation), so if you can limit your search to an updated range, it'll go faster!

Optimising the search system is completely different of course.  You're looking to optimise what gets indexed and read out during the search process.  Good use of labels can help again, but there is a part of it that jumps out - restrictions (on view) impose load on searches.  Again, it's not this simple, but if you imagine your search as a series of "narrowing down" searches, the first narrowing is "person can see space", the second "page matches word" and third "is page restricted".  The fewer things you can check for restrictions, the better.

All that said, your numbers look quite responsive to me!