posted on Mar 23, 2017 by Dominique Raviart
Tags: Tech Mahindra, Application Testing Management
We have reported several times on the service offering of the Performance Engineering unit within Tech Mahindra’s Testing Services business. Tech Mahindra Performance Engineering (TMPE) recently briefed NelsonHall about its work in positioning performance engineering within DevOps through its CONPASS platform.
Automating Performance Testing in the Context of DevOps
One big element of TMPE’s communication strategy has been to evangelize buy-side organizations about the benefits of introducing performance engineering early in the lifecycle. TMPE initiated a shift left approach, providing DevOps consulting services. There are clear benefits to introducing performance engineering early in the lifecycle. At the same time, developers also have to worry about functional requirements, and increasingly UX and security. To say the least, developers have a busy agenda at the software requirements level.
Another way of driving better application performance is through automation. TMPE has taken several approaches to automation, initially working around service virtualization and application release automation (both around CA software products, and then around IBM’s Green Hat). The company expanded to model-based testing (based on the Conformiq products).
CONPASS Platform
With the relative lack of client appetite for model-based testing, and the accelerated adoption of open source tools, TMPE has launched a new platform, Continuous Performance Assurance (CONPASS). CONPASS acknowledges Jenkins’ central role in CI/CD and includes plug-ins (e.g. NeoLoad and Dynatrace). The benefit of this approach is to put performance testing at the heart of CI/CD’s most widely used open source tool.
In a bit more detail, CONPASS has three elements:
- Test execution of the code on every build, at the development level. TMPE has, as part of CONPASS, its ePA IP that checks changes in the code and identifies the performance testing scripts that need to be changed. ePA integrates with open source software including QAtime, jProfiler, jUnit, jMeter
- Performance testing: through its PT IP, TMPE conducts end-to-end performance testing scenarios across networks, devices, and browsers, mostly during night shifts and at end of sprints. Tools supported by PT include HPE’s LoadRunner, Neotys’ Neoload, jMeter, and CA Blazemeter
- Production performance monitoring, though the integration of Dynatrace’s APM production monitoring COTS. An important element of the IP is the ability to track compliance to performance SLAs.
To date, TMPE has deployed CONPASS with one client, a telecom service provider in Asia Pacific. The client is migrating its OSS and BSS applications to AWS, taking a CI/CD approach, and commissioned TMPE to assess the performance of AWS’ IaaS offerings and select the ones with better performance. The organization is currently migrating five applications a month to AWS and TMPE is involved in the performance work. TMPE is also involved into three other deployments (with organizations in the banking and oil & gas industries).
CONPASS is provided as part of the service for short-term engagements, or under a licensing model for multi-year contracts. Implementation time varies: TMPE highlights that for clients who have implemented Jenkins and have additional DevOps, the deployment of CONPASS takes one week of technology assessment and three days for CONPASS configuration.
Looking ahead, TMPE wants to enhance CONPASS by developing/adding plug-ins to other CI/CD open source tools (Bamboo) and other production monitoring tools (AppDynamics).
Bottom Line
Tech Mahindra Performance Engineering is systematically exploring ways to automate performance testing in agile and DevOps scenarios in a true ‘trial and error’ approach. This is showing how agile and DevOps are rapidly becoming pervasive across the full software development lifecycle.
This time, TMPE has moved a step forward by creating its own testing performance platform. Platforms are becoming the new differentiation factor for any software testing service offering, just like testing frameworks were ten years ago. Welcome to the new testing.