Virtualization Track at Cloud Expo

Any die-hard software geek tends to be fascinated with solving the biggest, most complex IT problems enterprises face. In terms of testers, this means moving away from manual testing and getting teams to "go agile" with early development testing, or looking "between the boxes" into validating complex middle-tier technologies. In today's enterprise, most of the business logic happens deeper within the applications where services communicate transactions through an ESB and other messaging and data layers, both within the company's IT environment and with third-party systems such as Cloud-and SaaS-based services.

In these environments, concerns over manual testing can often get left behind as it is hard to validate a multi-tier SOA or BPMS-based app by testing at the user interface. Root causes of software issues are difficult to diagnose at the UI layer of modern apps. Still, we must certainly test systems from the user's point of view to ensure reliable outcomes in production - and that includes real people manually testing a UI. How can we eliminate constraints to make the manual testing process more productive as well? Virtualization of software behaviors provides a way to lift the productivity of manual testers as well as back-end integration developers.

Challenges of Constrained Manual Testing
We've all been there. We've got a team ready to start testing, our scripts are cued up, the lab time has been scheduled, and all of our test data has been loaded onto the system and verified. We start our run - and just a few minutes into banging on the system, testers start seeing hundreds of errors show up.

What Happened Here?
In today's complex system environments, it's not uncommon that the application we're working on has dozens, if not hundreds, of back-end dependencies. These dependencies span the gamut from simple and straightforward things like databases, LDAP servers, and web servers, to complex and arcane items like composite application services, mediation platforms, web services and external data feeds. More often than not, the majority of these dependencies is out of our control and has a lifecycle that is independent of the front-end web app that is being manually tested.

In essence, because of modern application complexity and service interconnectivity, manual testers can get stymied by the same factors that keep more technical integration and performance testers up at night:

• Lack of availability or limited access to needed back-end resources behind the UI
• Volatile, changing data that invalidates the testing efforts
• High infrastructure cost and effort of setting up and provisioning ready test environments

VSE and Automation for Manual Testers
Service Virtualization (SV) of external dependencies is an efficient way to mitigate the challenges associated with testing applications that have many dependencies. To date, this practice has normally been associated with automated regression, integration, and load testing and performance teams, but it is also applicable to the manual tester. Service Virtualization is the next step beyond hardware and OS types of virtualization, as it seeks ways to improve productivity by virtualizing the behavior and responsiveness of middle-tier components that are not suited for replication on a VM.

For instance, you might effectively virtualize a given Windows OS desktop or app server configuration on a VM. However, some items like a mainframe, a service under development, or a hosted cloud-based SaaS application are either unavailable or too large and distributed to replicate on a VM. This is where Service Virtualization can be brought into play to simulate the underlying system dependencies for testers.

Everyone tends to think in terms of tools for automated testers but, at the end of the day, virtualization is about enhancing productivity across the entire team, be it manual or automated execution. Organizations are starting to recognize that there's also tremendous uplift for an even larger audience of manual testers through SV.

Compare the roles using automation versus manual testing. If test automation is in play, there is a small team of highly skilled automators who are doing that work. When systems are not available they are inconvenienced, but, more often than not, their time is constrained and they may have a backlog of many different projects that they are working on. These skilled resources lose productivity due to the change in context, but there are further inefficiencies occurring in acceptance testing.

Manual testing is a very different game from automated testing in that manual testers usually consist of two different teams or audiences. There's the group of dedicated testers who focus in on the day-to-day validation of the application. But for larger test events such as Workflow Integration Testing (WIT), User Acceptance Testing (UAT), or End to End (E2E) testing a large number of business users are also brought into play. When a system dependency is unavailable then, it's not just a few people who are idle but sometimes literally dozens or hundreds of people.

Service Virtualization enables manual testers to move forward with an efficient and repeatable test flow, so they can expect to find consistent test data and full-time availability of application dependencies that exist under the UI they are testing. Furthermore, by using virtualization, they can reset their test data "on the fly." When they find they have made a mistake in their test workflow, they don't have to clear out or rebuild the test data in order to try to repeat a test.

Think about the number of resources involved in manual or "acceptance" testing within your enterprise, and the amount of cost and schedule coordination (and possible travel) required to get them ready to test software. While automation is a great goal to strive for, we know that there will always be a need for thorough testing from a user perspective. In these environments, Service Virtualization can be just the ticket for maximizing availability and minimizing downtime.