The client
SacSewer collects wastewater from a large share of Sacramento-area homes and businesses, using IBM Maximo to predict equipment problems across its collection system. With an upgrade to Maximo 7.6.1.x on the table, the district — working with its implementation partner, Maven — needed a performance baseline solid enough to bet a go-live on.
The engagement
Working together, SacSewer and Maven configured real user journeys to emulate the district's high-volume activities and likely bottlenecks — including GIS map usage through Maximo Spatial, the kind of heavyweight, real-world workload synthetic HTTP scripts don't reproduce. The test programme simulated up to 300 concurrent connections over a 24-hour window, generating reports that identified performance delays for the upgrade decision.
The moment that proves the method
Mid-project, pandemic-era hardware shortages kept pushing delivery of the intended physical servers, so a fallback was examined: host Maximo on virtual machines.
The standard test data said the VMs were fine. Dashboards green, metrics healthy.
The users said otherwise. Because real users were logged into Maximo during the test runs, their experience — slow, inconsistent, unacceptable — landed alongside the clean-looking numbers. That contradiction forced a deeper drill into resource-usage metrics, the bottleneck was found, and the VM route was called as a no-go — before it became a production incident.
When the physical hardware arrived and testing was repeated, the go-live decision was made with confidence on three green lights at once: the user team's sign-off, healthy performance data, and clean resource metrics.
The result
Analysis of historical Maximo activity let the team build accurate usage emulation — load that behaved like SacSewer's actual operation. That trustworthiness is what made both decisions possible: the confident "not yet" in 2022, and the confident "go" in 2023. A performance test that can tell you no is the one worth paying for.
Why we tell this story
This is the emulation-versus-simulation argument, lived: synthetic metrics alone would have green-lit an environment real users found unusable. Emulated real journeys — with real eyes on the result — caught it. And it's a partner-delivered engagement: Maven owned the client relationship and the programme; MaxTAF was the engine inside it.
Preparing an upgrade go-live? How our performance testing works → · Delivering one for a client? Partner with us →