Performance & Scalability Report
Find the limit before launch day.
Most teams discover their platform's limit in production, at the worst possible moment. We load tested Epsilon3 at program scale so you can see where ours is before you sign.
- How we generate load, and how test load maps to real users
- Results on a production-scale dataset, not an empty demo workspace
- 16 questions to ask any execution vendor, including us
Free report
Get the full report
Sign up to read the full report.
Prefer to talk it through? Book a demo.
Why we publish our numbers
Most vendors publish nothing.
Three years into a program you're carrying tens of thousands of procedures, millions of steps and hundreds of thousands of attachments. The platform has to hold when everyone is on it at once.
No published results
Vendors in this category talk about performance and scale, then publish no test results, no data volumes and no uptime history.
Tested empty
A fast demo on an empty workspace says little about how the system behaves on year-three data.
Tested once
A benchmark run years ago says nothing about this year's release.
What we tested against
Source: Epsilon3 full-scale load test. Step count estimated from dataset averages; other figures are direct counts.
What's inside
The method. The results. The limits.
Written for the engineers and program leads who will be asked to sign off on a platform before it carries a program.
How we load test
How 500 heavy-use load generators map to roughly 5,000 concurrent users, and why the harness never competes with the platform for resources.
Responsive under load
What happened during authoring, review and live execution at peak, with every part, tool and attachment in place.
Where the headroom is
Which component reached its limit first, and how we handled it partway through the test.
Built to scale out
The architecture choices behind the results, and the recent releases tuned for large libraries and long procedures.
30,000 users and beyond
How large organizations structure sites and programs so one site's heavy day can't slow another.
Testing that keeps pace
How the dataset and simulated load grow with customer workspaces, and what we will publish next.
A preview
Ask every vendor. Including us.
The report closes with 16 questions that separate published evidence from performance claims. Here are three.
- 01
Can I see your uptime history right now, without going through sales?
Ours is public at status.epsilon3.io, broken out by production environment, including FedRAMP.
- 02
How much data was in your test environment?
Ask for counts of procedures, steps, runs and attachments, not a fast demo on an empty workspace.
- 03
How often do you retest, and does the test grow as customers grow?
Ours runs weekly, with a full-scale load test every month at increasing scale.
Check the numbers yourself.
Read the full report, or bring a procedure you actually run. We'll show you it running in Epsilon3.