A Toast to Service Virtualization

kellyemo — Mon, 17 Aug 2009 22:37:00 GMT

"So a performance tester and a functional tester walk into a bar and the bartender says... why the long schedule?"

So you want to build out composite applications, deliver against business requests by assembling sets of pre-built services rather than build from scratch or orchestrate dynamic business processes from shared SOA and REST services driven by events? Does this sound like the nirvana of modern applications?

Architects get excited about realizing the promise of SOA, shared services and interoperable, loosely coupled application modules with standard interfaces driven by well-accepted policies. Developers get excited about building things using the latest and greatest standards. But, what about testing teams?

Functional testers need to test the functionality of their applications or conduct regression tests when the application under test isn't completely in their control or even in their domain of expertise. Now, they must test something that makes a call to or relies on a response from a dependent service or system that is virtually impossible to get access to. What if the dependent service is actually a mainframe application in production or a SAP system running a production business process, or even not built yet by another development team building out a shared service?

To date, the answer has been wait. Wait for your small time slice of the mainframe, wait until operations gives you a 2:00 am window, next Monday, (a business holiday) to access SAP, or until the other development team gets a stable enough build of their service for you to test against.

Performance testers need to test against a goal SLA that the business expects but their SLA is the aggregate of the performance of their composite application and all the downstream services that are part of the composition. How do they do that? What if they cannot even get enough cycles of a downstream system to prove performance? (let's say the downstream system is a third party SaaS application and each transaction costs the testing team major bank? It would be way too costly to run a major load test against the SaaS back-end). So, once again the answer is wait, or worst-case, take a SWAG (and we know what SWAG stands for..)

It's time to think about adding capability for testing teams to virtualize their constraints and dependencies so they can continue testing throughout the application lifecycle. And it's time to stop expecting testers to become experts in challenging back-end systems such as mainframe protocols, integration systems such as MQSeries or JMS, or proprietary programming models. Enter the role for Service Virtualization.

In case you missed it during summer vacation, iTKO announced that HP will be reselling iTKO LISA Virtualize. The waiting is over. Service Virtualization will help testing teams keep forward progress in the face of an array of constraints....

Now that's worth raising a glass!

Making Sense of HP Software Universe -- A Look Back at Last Week's Event

kellyemo — Fri, 26 Jun 2009 18:00:00 GMT

Last week, I spent a whirlwind four days at HP Software Universe, HP's bi-annual customer and user community event for all things software. Between the content-rich interactions with customers, attending and managing sessions, and informal meetings with partners and colleagues, not to mention dinners and excellent wine (can we say "food coma"), it was a productive week but has taken me at least that to recover... thus the belated blogs this month.

That said, I don't want to bore you all with my chronological review of a week in Vegas. (no I did not gamble and take home a jackpot or lose it all and sound like a country ballad - lost my house, car, dog.. etc), rather, I'll just share a few insights that I heard and what it means to making sense of SOA.

1. Insight #1 - SOA is not dead. Have we heard enough about this already? The debate continues unabated in cyberspace. That said, what I saw at HPSU was SOA everywhere yet SOA never upstage. Our partners showing testing tools, management solutions and best practices were talking services, composite applications, loosely coupled architectures - all things SOA as just core to what it is they do.

2 Insight #2 - Modern application styles are starting to really impact quality and testing teams and there are not a lot of proven best practices and success stories yet. We are at the "now we get it, we need to test inter-related project components in the context of a composite application, services that are used/consumed in multiple places but that don't have a UI, orchestrated business processes, and we are not sure how" phase. QA teams now have to take the reality of interdependent lifecycles, shared services and intermediaries seriously. Combine this with the desire to align development and QA closer together, embrace agile methodologies for testing earlier in the cycle, and in general deliver applications faster, better and cheaper, and QA is elevated to a lead role-a critical player in the entire lifecycle of delivering, changing and evolving composite applications.

The good news is that there are technologies to help; such as testing suites designed for the concept of shared, dependent and often "headless" services, and virtualization capabilities to enable testing services and interfaces that have dependencies on scarce resources (can we say mainframe?) without having to actually have the physical resource available. In fact, one buzz of HPSU was the interaction of HP service-oriented functional and performance testing solutions with ITKO's service virtualization solutions.

3. Insight #3 - Customers are doing governance and it's working... just not all at once, nor should it. For some reason, of late, governance has been getting a bit of a bad rap. I had more than one person ask me, should we even call it governance since it's a "dirty word"? I think the image of total lock-down, loss of flexibility and boiling the ocean comes to mind.

Reality couldn't be further from the truth and that's the message that vendors and customers need to get out in the industry as fast as possible to avoid a retreat back to the wild-west state of every development team just doing it's own thing or trying to herd cats using a combination of wiki's and spreadsheets.

What I saw at HP Software Universe were pragmatic implementations of SOA and more broadly, Application Governance that were working and returning real measurable value. I saw customers solving specific, well understood and agreed upon challenges with a combination of ownership, applied best practice, well-defined policy, and a flexible approach.

For instance, one customer is using HP's SOA Systinet solution to define the specific stages of a lifecycle of modernizing aspects of a major application overhaul. In the lifecycle definition, owners are clearly defined and a chronology of decisions regarding when the new version of the application component is ready to go live is not only documented but guided via software. Policies are applied to certain actions but not too many and what policies to apply have been determined via a key stakeholder voting process so that all teams who are impacted by the policy know why and what the rational is for the policy. For this customer, policies are mainly focused on design-time. They have run-time policy management as a goal but will begin that when the organization is ready.

I heard from another customer who is not ready to use governance yet to automate lifecycle hand-offs but who is effectively using it as a system of record and to go through a review process for what services to implement as shared services - the service candidate phase.

Finally, I heard from a customer who is using governance to document the whole "who, why, how and what would happen if we didn't do this" tribal knowledge aspect of policies. They use governance processes to thoroughly assess the top areas of pain, determine if they can be fixed through governance and the application of policy, have a vetting process to prioritize what to govern with policy, document all assumptions including clearly telling a story as to what is the negative outcome if the policy isn't followed, and then if the organization votes that this is a high priority policy to implement, uses the governance process to put the policy in action, and measure and tune the outcome over time.

In none of these examples, do we see "boil the ocean". It's pragmatic governance with a specific set of vetted organizational expectations being measured and delivering results.

So what is my take away from HP Software Universe in terms of making sense of SOA? SOA is here to stay, and everyone is trying to figure out how best to keep IT humming with the major changes SOA brings. There are real examples of best practices, new processes and the application of technology to make IT more successful in the modern application world and people will line up to hear them. If we are to keep the pace of modernization accelerating, we need to tell more of these anecdotes with less fluff and more real practice. And in the words of Paul Harvey, "and that's the rest of the story"

More to come on making sense of SOA...

Making Sense of SOA Blog : Service Virtualization

A Toast to Service Virtualization

Making Sense of HP Software Universe -- A Look Back at Last Week's Event