Simulations in SDLC: Billions of dollars saver
By Cat Yong
Having to do more and faster with lesser resources, is an IT issue that has been around for quite some time. The reality also is that in the application and software world, applications that fail when customers use them, will hurt customer experience, and that is a big problem, for any business in any sector.
CA Technologies’ VP of Research and Development for Application Delivery/DevOps, Ruston Vickers pointed out, “When we look at businesses and how they deliver software today, they are having to move more quickly.”
The way that applications have always been developed; manufacturing conveyor belt style; does not cut it anymore, and in fact, perpetuates fundamental issues about the software development lifecycle (SDLC).
CA’s Application Delivery business unit is a new business unit which tries to alleviate these issues by focusing on Service Virtualisation, Continuous Delivery System, Complete Monitoring and Collaboration Enabling Technology
CA Technologies’ SVP of Global Presales, Trevor Bunker said, “The way that IT departments build apps, is the same way that most manufacturing facilities build purses, for example.”
It is a long and arduous linear and serial process that involves processes like requirements elicitation, planning, build process, test and release. In contrast, once applications go into production or operations mode, processes become continuous.
|(L-R): PF Vilquin and Trevor Bunker. You might be able to tell they really enjoy their work|
CIOs find that these two ‘environments’, Dev and Ops, have two fundamentally different speeds and separate cultures, when actually they could and should be in sync and collaborating.
Enter CA’s LIS, an application delivery suite with tools that DevOps methodologies require.
“How to go faster (with application releases) when the conveyor belt only moves at one speed?” Bunker posed the hypothetical question that would also answer how to build applications whilst getting testing automated with rollbacks that are as easy as deployment.
CA’s collaboration enabling technology for example, potentially can allow operations technology to enter into the development space as smoothly as possible.
Service virtualisation – Star Trek-type of tech
Currently, there are too many ‘dependencies’ to be delivered before a piece of software or app could move forward to the next phase in its development lifecycle.
What service virtualisation does is simulate or virtualise processes, negating need for the lifecycle to continue to be serial. It enables processes to run in parallel, which inevitably moves development along faster.
Bunker described, “If you want to get a new mobile payment platform out, for example, you could in parallel have all the different components started at once ie. user interface, backend integration, back office applications and so on.”
PF Vilquin, CA Technologies’ CTO in Asia South pointed out that having a front-end user interface developed earlier on in the process, means bug or error detection could also happen earlier. “Typically software is tested at UAT (user acceptance test) stage, but that is only when the UI is ready.” Most times that is too late.
Bunker shared that in Malaysia, a big issue is that no matter what industry an organisation is in, almost always no one has enough money for software testing.
Banking for example, needs an almost exact copy of the core banking platform and a certain amount of infrastructure to support testing. “But companies can’t afford it. They can’t recreate the same environment to same scale.”
With service virtualisation, CA was able to help a large bank in US, to save over US$30 million by avoiding having to stand up another dedicated environment with a full copy of their banking system.
Vilquin explained, “Service virtualisation isolates components and make them behave as how they would behave if they were in production. Developers can test immediately in full environment without having to make assumptions about how the application would work.
“So all the different components can all be in their own environments, believing they are talking with the backend systems (and be simultaneously tested), instead of having to wait for a component before it that they rely on, to be ready to progress forward.”
A goal that testing could try to attain is 60% detection of bugs/errors in development, 20% of defects caught in testing and final 20% caught in UAT. “Problem is, no faults are detected at all in development mode!”
There are short windows of time to get an app into production, and when done manually it could take fifty people overnight at the lab to go through all the paces of releasing a software. Automating its release is faster and can reduce human error.
“It maximises chance for successful deployment of application” said Vilquin.
What do CA Technologies’ solutions bring to the table?
Significantly, CA Technologies boasts first mover advantage – being first to go to market with ITKO LISA solution which later became CA LISA after acquisition by CA Technologies.
Bunker said, “That means we have greatest maturity and experience working with customers.”
Bunker also believed that being a leader in application performance management (APM) stands CA in good stead, as they would have a huge repository of performance and transaction data, collected over the years.
“Big corporations are recognising that they can take the data we have collected and start building models/simulations.
“So, CA has unique capability to link to production for wealth of information. The more realistic you can make the simulations, the better the app will perform in production mode,” said Bunker.