What is the performance impact of Git Integration for Jira plugin?

[Pavel Boev]

We are considering to use Git Integration for Jira add-on on our Jira Software Server instance https://marketplace.atlassian.com/apps/4984/git-integration-for-jira?hosting=server&tab=overview

The plug-in adds a lot of functionality to Jira (like repository browser, comparison of files and branches, etc.), but I am worried about performance impact.

We will have about 80 Gitlab repositories indexed by the plug-in

In addition we have Crucible, which is already used for ~500 repositories.

Here are my questions:

• What is the impact on the load time of View Issue page, when there's no referenced Git commit? In our case more issues will be linked to commits in Crucible, compared to issue linked to repos, indexed by the plug-in.
• What is the extra storage that we must prepare for the add-on (file system and DB server)?
• What is the average increase of CPU load on an instance with this plug-in (indexing 80 repos), compared to instance without this plug-in?
• Is this plug-in more resource intensive than the regular Jira DVCS connector plug-in (it can be tricked to work with Gitlab EE)?

Thanks in advance

[Mark L. Smith · Accepted solution]

Hi Pavel,

What is the impact on the load time of View Issue page, when there's no referenced Git commit? In our case more issues will be linked to commits in Crucible, compared to issue linked to repos, indexed by the plug-in.

No referenced commits is almost no additional load time.

What is the extra storage that we must prepare for the add-on (file system and DB server)?

DB usage is unremarkable --- approximately 10s of bytes per Git commit.

Local storage will be a function of the total size of all bare repo clones plus an indexing factor. The indexing factor is related to number of commits --- not size of the repos.

What is the average increase of CPU load on an instance with this plug-in (indexing 80 repos), compared to instance without this plug-in?

We do a lot of work around indexing and load performance. However, 80 repos is smaller than the typical customer install (on the high end we have customers with 30,000 repos).

We have a lot of settings to allow an admin to enable/disable parts of the plugin to manage performance as well.

However after the initial index and especially for 80 repos, I don't think you'd see a negligible increase of CPU with the stock settings.

Is this plug-in more resource intensive than the regular Jira DVCS connector plug-in (it can be tricked to work with Gitlab EE)?

They're different products so the comparison gets hard. This plugin does a lot more than DVCS so if you use the "more" then I'd expect the resource utilization to be less. If you disabled the "more function" --- then it might be about the same.