Integrate with Jira - api limits

Question

We're a service provider that has a lot of automations against tickets in our current ticketing system (Freshdesk). We automatically create and update tickets via the Freshdesk APIs and are having some problems with our current API limit of 20000 requests per hour. We have also encountered some bugs in Freshdesk that throttles our requests even when we're nowhere near the hourly api limit.

Anybody else have experience with integrating with Jira from an API and rate limit perspective? There appears to be no documented request limits for Jira. We anticipate needing around 20000 requests per hour (or the equivalent of 333 req per minute) but want to choose a solution we can grow with over time.

Want to hear from the community any good or bad experiences with building advanced automations/integrations against Jira API.

Danyal Iqbal · Answer

The support team at Atlassian should be able to help you out with Cloud.

Keen Fann · Answer

Thanks for the feedback, appreciate it! I agree that running Jira in our own VM:s would solve this particular challenge. That said, we prefer to avoid VM:s if possible. I wasn't clear in my original post but I'm primarily interested in Jira Cloud. :)

Danyal Iqbal · Answer

That is a very interesting setup. All i can say is that it really depends on how long each request takes and the sizing parameters of the environment. if a single request takes 1s or more than you would need to bump up the hardware .

The tomcat Server can entertain 200 simultaneous by default . You can increase the maxconnection and maxthread parameter in the server.xml to optimize for performance( you would definitely need more memory and CPU resources.. think 32 GB RAM and 12Core for 20k request/s with single request time~500ms ). Also have a look at the sephamore value to optimize worker threads as well.

Response time will increase under load!

If you are not doing anything crazy with the rest api tomcat can easily handle 333 request per minute.  If you are running jql queries with large result sets or other time consuming operations than I would start thinking about cluster setup and load balancing approaches.