Okay, team. I've been running a real-world stress test on Jira for the last six months with a 150+ dev org, and the results are... not great for velocity. We're talking about a project with 15,000+ issues, 50+ active epics, and complex branching automations.
The core issue isn't feature bloat—it's **architectural**. When your project scales, every interaction feels like it's querying a denormalized database without proper indices. Simple actions become slow HTTP requests chaining dozens of underlying API calls. I instrumented some basic operations:
```javascript
// Simulating a dashboard load for a 'board' with 500 issues
1. Fetch board configuration (fast)
2. Fetch filter for board (slow - complex JQL)
3. Fetch issues page (very slow - JQL + rendering)
4. Fetch custom field values (multiplier per issue)
5. Calculate aggregate stats (in-app, blocks UI)
```
Each step adds latency, and it's multiplicative. Atlassian's official response typically points to:
* **"Optimize your JQL filters"** (We did. It helped maybe 10%.)
* **"Check your add-ons"** (We run barebones.)
* **"Consider Data Center for performance"** (The upsell is real.)
What we actually observed:
* **Page load times** for complex boards: 12-18 seconds.
* **Simple issue creation** via UI: 5-7 seconds.
* **Bulk operations**? Forget it. Timeouts are common.
So, what are the alternatives that *do* scale under similar load? I've been testing:
* **Linear** 🚀 - Blazing fast. GraphQL API, real-time sync. Struggles with *extremely* complex workflows but perfect for product/dev alignment. Lacks some enterprise audit features.
* **Shortcut** (formerly Clubhouse) - Feels like "Jira but rebuilt." Clean API, sensible defaults. Handles large backlogs well. Their batch operations are actually reliable.
* **Height** - Combines projects *and* docs. Async-first. Surprisingly performant with 10k+ tasks. Automation is code-based (TypeScript), which is a pro for us devs.
The pattern? Modern alternatives use a single GraphQL/REST query where Jira might make 10+. They treat "bulk" as a first-class citizen, not an afterthought.
Has anyone else done a deep dive on Jira performance at scale? What was your breaking point, and what did you switch to? I'm especially curious about data migration war stories.
--experiment
Prompt engineering is the new debugging.