Performance testing
A load test is only as honest as the data behind it. Production has the volume but can't leave production; a masked subset is safe but small. Synthesized gives performance systems production-scale, masked data, then generates it up to the volumes your next peak, rollout or acquisition will bring.
On SAP, we measure results on your own data in a 10-day validation.
Why tests mislead
01
02
03
04How it works
Volumes, distributions and skew for the objects your scenarios touch, measured inside your environment.
Subset the business processes in scope and mask names, addresses, bank details and salaries.
Scale customers, orders and postings to the target, keeping relationships, distributions and the long tail.
Write the data set into the performance system, run your load test, and rebuild the same data set for the next release.
Compare
| Full production copy | Masked subset | Masked and scaled up | |
|---|---|---|---|
| Volume | Today's production | A fraction of production | Today's, or the target you set |
| Distributions and skew | Real | Often lost | Kept from production |
| Personal data | Every real record | Masked | Masked, plus records with no real person behind them |
| Next year's volume | Not available | Not available | Generated to target |
| Repeatable | A copy project each time | Yes | Yes, same data set each release |
When to use it
S/4HANA migrationMigration cutoverProve batch jobs, conversions and the first month-end on S/4HANA before go-live, at full volume.Migration test data
Business peaksPeak and closeMonth-end and year-end close, peak order intake and payment runs, at the volumes those days really bring.
GrowthRollouts and growthNew countries, plants or acquired businesses add volume. Generate it before it arrives.
Upgrade testingUpgradesCompare response times before and after a release, on the same data set each time.Upgrade testingWe've got you covered
A copy gives you today's volume but brings every real customer and employee into a system with wide access, and it can't show next year's volume. Masked, scaled-up data gives the volume without the exposure.
It keeps the relationships, distributions and skew measured in production, such as a few very large customers and long document chains, so queries and batch jobs meet the same shape of data.
Teams have grown test data 200 times, from 100,000 to 20 million entries, in a deployment outside SAP. You set the target per object, such as twice today's orders.
Any. The data lands in the SAP performance system, so LoadRunner, Tricentis NeoLoad, JMeter or your own scripts run against it as they do today.
Yes. Load the converted volumes, or the volumes you expect after go-live, and run cutover jobs and the first close before the real cutover.
Next step
Tell us which scenario worries you most. In a 10-day validation, we build a masked, scaled-up data set for it on your system.
SAP, S/4HANA, SAP HANA, SuccessFactors, Ariba, Concur and other SAP products and services mentioned herein, as well as their respective logos, are trademarks or registered trademarks of SAP SE (or an SAP affiliate company) in Germany and other countries. All other product and service names mentioned are the trademarks of their respective companies.