Case Study · Energy / Utilities · 1 year
One of the largest energy companies in U.S.
Re-engineered storm-scenario performance testing for the Outage Management System using a Virtual Table Server model — cutting virtual users from 17,750 to 2,500 while delivering true 24-hour storm simulations and higher confidence in OMS stability.
The challenges
- Excessive LoadRunner license costs due to inefficient test design — stacking virtual users across 24 hourly groups drove virtual-user requirements up to 17,750.
- Difficulty realistically simulating highly variable storm conditions across channels like IVR, web, mobile, and smart meters, while preserving timing relationships (e.g., the 3–4 hour delay between Power Outage and Power Restoration Notifications).
- Need to validate Outage Management System (OMS) resiliency under extreme, real-world conditions after significant issues during Hurricane Isaias.
How we solved it
We replaced the multi-group, hour-by-hour approach with a single-group model per script driven by the Virtual Table Server (VTS). Instead of ramping users up and down every hour, each script runs with a fixed number of virtual users for the entire test, with variability coming from timestamped transactions in VTS rather than from adding more users.
Each VTS record represents a real transaction with its type, customer data, and exact execution timestamp. Virtual users pull records and wait until the scheduled time before executing, so traffic volume is controlled by how many records exist for each hour — not by how many users are running. This preserved natural peaks, valleys, and delayed transaction patterns.
The new VTS-based model enabled long-duration, production-realistic testing: millions of timestamped events preloaded into VTS, the same group of virtual users running continuously for 24 hours, and load patterns matching real storm behavior — making testing repeatable, scalable, and cost-effective.
Results
- License usage dropped from 17,750 to 2,500 VUsers — an 85% reduction (15,250 VUsers saved).
- Infrastructure cost reduction: fewer load generators and lower CPU, memory, and network usage during tests.
- Operational efficiency: simpler single-group configuration, faster setup and execution, less troubleshooting.
- Risk avoidance: higher-fidelity storm simulations reduce the likelihood of production outages — preventing incidents during major storms can save millions in outage response and customer impact costs.
Expertise provided
25 years of LoadRunner (performance testing) experience, 10 years of C programming, and previous experience with other energy companies resolving the same issue. Application under test: the Outage Management System (OMS) on SAP.
Solutions used
- OpenText LoadRunner Enterprise (LRE)
- OpenText Virtual Table Server (VTS)
- MS Excel