Our current setup runs Jira on a 4-core 32-bit RHEL machine with 4GB of RAM and everything works fine right now. But we expect more users soon and want to figure out how many concurrent users our Tomcat server can support when running Jira.
The Jira instance is running on a dedicated Tomcat server that doesn’t host any other applications. So any performance data about Tomcat alone would be useful too.
We get about 6000 page views per day on average. Has anyone tested similar configurations? Would love to hear about your experience with user loads and performance.
Based on my experience, your current configuration may face limitations sooner rather than later, primarily due to the 32-bit architecture which restricts the JVM heap to approximately 3.5GB. This limitation could hinder performance when dealing with larger datasets or intricate workflows in Jira. I’ve observed that setups like yours can typically support around 50-80 concurrent users, although performance may degrade beyond that threshold. Given your average of 6000 page views daily, you’re likely seeing about 20-30 concurrent users during peak times, which is manageable now. However, keep in mind that user activity matters as much as user count. Activities like running detailed JQL queries, heavy dashboards, or large file uploads can quickly deplete resources. I recommend closely monitoring your JVM’s memory utilization and considering a transition to a 64-bit setup to allow for a larger heap as your user base expands.
Honestly, ur setup sounds just like ours before we upgraded. That 32-bit limitation will def bite u soon - Oscar’s right about that. We started seeing slowdowns around 40-50 concurrent users, but it really depends how heavy ur workflows r. Tomcat can handle way more, but Jira gets resource-hungry fast. I’d start planning that 64-bit migration now before things get painful.
We hit the same bottlenecks last year. The 4GB RAM is your real problem - CPU cores aren’t the issue. Our monitoring showed Jira eats 2-3GB just for basic stuff, so you’ve got almost nothing left for peak times. With 6000 daily page views, you’re probably seeing 25-35 concurrent users during business hours. That’s manageable now, but here’s the thing - performance doesn’t drop gradually. It stays fine until you hit a wall, then tanks hard. Database queries and custom field calculations kill performance way more than page loads. Tune your JVM garbage collection and watch response times like a hawk. We bumped up to 8GB RAM on 64-bit and saw massive improvements in both speed and user capacity.