Can JSD (DC) handle millions of issues a year?

[David Sokolowski]

I'm working with a client that is replacing their proprietary call center issue tracking system with Jira Service Desk (Data Center). We are all very excited to get them running on this great tool, and I've got them setup with their Calls as standard issues, and their Subjects as sub-tasks (average of 3 Subjects per Call).

Getting into the scale of their call center has thrown me off a bit. Between both the Calls and Subjects (issues and sub-tasks), they're going to be creating between 10-12k issues a day. That scales up to around 250k issues a month, for about 3 million issues a year.

Since I imagine they want to continue using the tool for a while, they're looking to crack the 10 million mark early into their fourth year. Now from a software angle, I think they're heading down the right path by choosing Data Center, and I believe they have the hardware to handle this scope. 

Here are some other facts: They add few attachments; there are about a dozen simple fields; they will be using nFeed to call the CRM for customer data (via JSON API); there are only 150 users at any given time; they are not using boards but will be searching and querying regularly; they look to organize their instance into about 8 projects; the workflow has about six steps (though most will just use Open > Closed). 

What other considerations need to be taken into account for a Jira instance of this scale? Do you think Jira can handle this many issues, even for the overall light-touch system that we're creating? I know a lot depends on the scale of the JSD configuration, but: 3 million issues a year is a lot.

[Mirek]

Maybe number of issues is not a problem but all actions around issues.. yes, it is! You can have 1 milion issues but only 1 single active user and this will work smoothly but when your real time users number increase and those users have their own dashboards, searches, subscriptions, boards, add, remove, .. just be active and use system as usual then it is starting to make things slow.

Some time ago we did with a couple colleagues a JIRA test to answer a question how much issues JIRA can handle? .. We had very powerful machine with many CPU cores, more than 32GB machine etc. and created 18 milion issues (and creating was not a problem) .. what happen after that? ... for example .. Dashboard opened around 20 min :) . Operations were possible but that is not how JIRA should be used. And we only used few active users at the same time. I do not even say how long a reindex took :) ..

So one node is not good option if you want to go with "milions" of issues. How this would be working on multiple nodes? It is a open question but it still might be not super fast. Everything depends on how instance would be used and infrastructure behind. I recommend do not think about this now just scale when there is a need. Add resources and nodes and then react if you see any slowness. You can always archive some projects.. and maybe some day JIRA would have a cool performance improvements that do not exist now.