The ITKO Blog: Service Virtualization and Application Lifecycle Optimization - CA Technologies

Whitepaper: Demystifying Cloud-Based Testing and the Related Benefits

You might have noticed a lot of commentary about Cloud Computing recently on this blog, and it is not a coincidence. In our conversations with analysts, peers and of course customers in the field, we are frequently asked about how this approach can add value to the testing and virtualization practices we support.

We have just put out an iTKO whitepaper on this topic, titled "Demystifying Cloud-Based Testing and the Related Benefits." The PDF is free to download.

There are many hard lessons and new techniques learned from industry-wide attempts to decouple and flexibly assemble enterprise software around key business processes. In order to reduce risk as we move into Cloud environments, we need systematic testing that gets around the roadblocks of inaccessibility and unexpected costs. We need to make use of automation and virtualization of the extended environment as much as possible. Cloud-based services need to be governed and validated as part of a continuous business process.

At this stage of the Cloud's development, there are many vendors and service providers that can contribute value to a comprehensive deployment and development environment solution. This short paper does not seek to recommend a comprehensive solution set, rather it focuses on gaining value within the software testing lifecycle. It outlines the key challenges and the top 4 strategies in play - first steps for virtualizing and testing distributed assets in the Cloud. We hope you find the information useful, and invite your feedback.

Testing Cloud Performance Strategies: Mike Kavis

Read a recent article from a very active pundit and practitioner in the Cloud Computing space, Mike Kavis, titled "Architecting the Cloud: Testing performance in your cloud application."

In the article, Mike advocates a layered approach to testing cloud-deployed applications that makes a lot of sense, in terms of the kinds of performance testing you can actually do while the design and development lifecycle is going on. To summarize, this involves testing from the system roots up while the solution is firmed up in this order:

1. Infrastructure. Test performance of the cloud provider(s) to ensure they can sustain the needed load.
2. Data. "it is time to analyze the impact of encryption, transformation, data replication, and the various ways that data is being manipulated to address security , compliance, reliability, and scalability requirements."
3. Business Logic. Interactions of services and processes.
4. User experience. Does the application meet user needs?

[Image Source: Mike Kavis]

This is a solid approach for testing to happen while a "build from scratch" cloud-based project is being designed/developed. Testing the raw performance level of the infrastructure first is something worth validating early on, and so on up the stack, finally testing the business logic and application when they come together.

Mike talks about how covering the core performance of the cloud infrastructure early can mitigate risk at deployment: "... It would be very expensive to find performance issues from the lower levels of the architecture at this point. That is why I recommend the layered approach to performance testing. What I also like about this approach is that you can start testing very early in the life cycle."

To expand on this discussion, one way to shorten this cycle is through Service Virtualization, which is the practice of simulating and modeling the behaviors and performance characteristics of components, business logic and even UIs in the cloud that are not yet available for live testing. This kind of virtualization can go a long way toward eliminating service and 3rd party system dependencies and bringing performance testing to bear much earlier in the process.

Then, after that rollout, there remains a continuous validation need. (Only so much you can cover in a blog, so this is a topic we'd like to see Mike address in a future post... por favor... : )

For instance, the infrastructure itself may behave differently under the kinds of data and usage variability in ways that you could never predict with the raw-traffic or user interface kinds of performance benchmarking. New versions of services can be introduced, the Cloud provider can do something that affects their performance -- the list of risks goes on and on. Continuously validating going forward is a way to ensure that there are no unintended consequences of change at any one of the above layers.

All in all, it is a nice development to see practitioners showing real architectures and methodology around Cloud Computing as it is being applied for the business, as opposed to continuously defining the space, so I'd recommend keeping up with future posts from Mike.

Podcast: Virtual Strategy Magazine on VSE Benefits

Virtual Strategy Magazine's Carryl Roy recently recorded a podcast briefing with iTKO about the new features of LISA 4.6, with a special emphasis on the Virtual Service Environment (VSE) capability of the product, and why a more expansive vision of Virtualization is needed for today's heterogeneous, ever-changing software environments.

Virtual Strategy: "iTKO Rolls with the Punches, Releases New Version"

Virtualization of hardware has already made great strides in reducing costs and increasing flexibility in the staging and deployment of software in the data center, but there is much more work to be done in optimizing the software lifecycle before production.

We are seeing companies pressured to keep IT budgets down, while the complexity and volume of data accounted for in applications keeps growing exponentially. There is a need to prune away many of the costly dependencies that can drag down the productivity of software development and testing teams. You can listen to or download the 15-minute podcast here.

Talk about Service Virtualization at HP Software Universe '09

At this year's HP Software Universe we're thrilled to be talking about a quality practice where we've seen incredible strides made over the last two years. The concept sounds simple enough: Virtualize away my upstream constraints and dependencies so that the test lab (for instance HP LoadRunner and Unified Functional Testing) solutions have a self-contained environment for testing.

However when you look at how complex and interdependent today's applications are, you begin to understand why replicating these complex behaviors -- such as unavailable or secured systems, third-party Services and components that are still under development -- has been resistant to replicating via conventional hardware virtualization means.

What is really cool is how we've seen customers of HP and iTKO bring this particular challenge under control. If you are planning on being in Las Vegas for this year's HPSU 2009 - June 15-19 - we invite you to take an hour to join HP Senior Product Manager Jeffrey Meyers, and iTKO Founder/Chief Geek John Michelsen to find out how these Virtual Test Environments can eliminate constraints (and costs) in a big way.

Session 1497: Thursday, June 18 - 10:30am - 11:30am
"Eliminate testing constraints and test earlier in the lifecycle: using service virtualization to drive testing momentum"

Here's the abstract of the session:
If your IT team is struggling with obstacles outside your organization, then join HP and iTKO and find out how HP testing solutions, combined with the iTKO LISA Virtualize platform, can minimize external dependencies and help you test earlier in the development cycle, before a complete test environment is ready to meet your testing objectives. You’ll hear how the combined HP and iTKO solution works in complex shared environments of modern applications, agile development techniques, and volatile data sources to make your testing efforts constraint free.

If you are planning on being there, look us up on Twitter as well: http://twitter.com/itko and drop us a note or feedback on your thoughts.

It seems in June we really catch the "last gasp" of interesting IT events, and this month is no exception - besides attending this excellent summit, we are bringing in the summer next week presenting sessions in Chicago at Technology Executives Club and a Federal roundtable at GTRA in Virginia. Hope we see you at one of these key events soon.

Is SOA Governance Becoming a Silo Unto Itself?

A colleague recently passed me an article from Joe McKendrick, "Is SOA governance also its own silo?" that really got me thinking. If SOA governance is a software development activity, does it become a self-defeating initiative?

In the post, Joe cites an Ovum analyst and blogger we've followed for some time, Tony Baer: "There are various issues at work with SOA governance silos, Tony said in the Ovum report. First, there's a perception that SOA-based technology is "reserved for elite practitioners." Of course, relegating SOA to an elite group "compounds the waste and duplication that SOAs were supposed to avoid."

The threat of Vendor-based SOA is well known - basically locking you into one technology stack isn't a SOA approach by definition. But neither is the reaction of having a SOA Governance strategy assembled strictly by a star chamber of developers within the company.

What is ironic here is that the very concept of SOA Governance is to accommodate federated authority, loose coupling and flexible assembly of technologies around business needs. If indeed the concepts and tools required for SOA Governance become so specialized or complex that only a certain class of technologists can use (or reuse) them, instead of the business subject matter experts, we are inherently missing the boat.

This has been one of the biggest questions we ask ourselves as an organization -- "Am I doing everything in my power to make ensuring quality and agility as approachable for the customer as possible?" We hope all vendors and services firms involved in SOA consider this line of reasoning. Specialist IT teams, if put into a [[SOA Governance Silo]] will be increasingly stretched and understaffed, plus the farther removed they are from the business, the more they will take a technology-first approach to governance (what are my tools, what WS* protocols, etc.), and limit the business value they can deliver over the long term.

Webinar 6/18: Automated ALM & Dynamic SQA with iTKO & Forrester's Visitacion

Wanted: A way to tie together the software delivery advantages of Application Lifecycle Management (ALM), with Dynamic SQA, which provides the in-depth quality coverage required for today's complex and constantly changing enterprise apps. What practices and tools are required to make this connection happen?

We are excited about an upcoming webinar on Integration Developer News that should amply address both. iTKO Founder/Chief Geek John Michelsen is joined by Forrester VP Margo Visitacion, a leading expert in ALM and Software Quality. Ensuring predictable results and on-time delivery of enterprise software requires testing, validation and virtualization across a myriad of changing technologies and services. Customer examples and current methodologies will be explored.

Here's the official description and signup link:

iTKO and Forrester Webinar 6/18: "Removing ALM Obstacles for Greater Agility, Cost Savings, and Lower Risk"
Experts from iTKO and Forrester Research present trends / technologies behind automated testing and validation to streamline ALM for higher productivity, accuracy and agility. Attend and Learn how Testing, Validation and Virtualization are being employed to support the best practices of Dynamic SQA (Software Quality Assurance). With with strong Dynamic SQA in place, enterprise IT teams are poised to deliver predictability, patterns and reuse to better manage complex software projects.

Our last webinar on SOA and Integration Efficiency (view archive) had more than 500 attendees, and lots of great questions we were happy to address (either on the session or afterward). We hope you will join us for this one!

US Federal Government Plays Leadership Role in IT Again

Our Federal government has played an innovative role in SOA governance (see Federated Computing demands SOA Governance & Testing) and now they taking a leadership role in distilling the value of Cloud Computing.

James Urquhart recently wrote, Are the feds the first to a common cloud definition? James wrote that the "public sector is eager to get rid of data centers in order to focus funds and effort on delivering value to their respective 'customers.' ... it appears that SaaS and collaborative technologies stand to benefit greatly from this trend."

The National Institute of Standards and Technology, a nonregulatory arm of the Commerce Department, has developed a draft definition for federal use of cloud computing. James quotes Reuven Cohen, in that "cloud computing will be the de facto standard definition that the entire US government will be given ... In creating this definition, NIST consulted extensively with the private sector including a wide range of vendors, consultants and industry pundits..."

Here is what they came up with: "Cloud computing is a pay-per-use model for enabling available, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort or service provider interaction. This cloud model promotes availability and is comprised of five key characteristics, three delivery models (SaaS, PaaS, and IaaS), and four deployment models."

This is just the opening to the definition and James provides the details on the five key characteristics, three delivery models, and four deployment models in his post. The definition statement acknowledges that it will evolve over time, but it tries to capture all of what is considered the cloud for now. This is a noble and useful effort. We have been writing a bit on one of these key characteristics, rapid elasticity. This can promote the cloud's use for application development and testing.

Thanks to the NIST for taking the lead here. I hope it can speed the use of the cloud within both defense and civilian branches of the government to maximize the efficiency of technology investments.

How to be Fast and Agile with Increasing Services Complexity Without Falling on your Face

Agile development is spreading in the SOA world to rapidly decrease development times, but if testing is not done properly, you might lose all the efficiency gains through having to conduct a massive cleanup effort afterwards. Like building bridges, fix it as you go is a much better approach than fix it later with massive traffic flowing over the structure. InfoWorld's Eric Knorr writes about this in his recent Modernizing IT article, "How to be agile without falling on your face."

Eric reports on a conversation with iTKO’s Derick Townsend. Derick said that companies within an M&A cyclone -- particularly in times of high volatility like we are seeing today in financial services markets -- are desperate to accelerate application development projects.

"They get 80 percent through a project, and they think the end is in sight, but the last 20 percent really stymies them, and that has to do with the test and delivery part. They're dealing with so much complexity. Telcos are experiencing this too. The components and underlying systems are changing constantly underneath," Townsend said.

Eric goes on to write about how iTKO provides what we call Service Virtualization, which is different than traditional hardware virtualization. Instead, iTKO LISA absorbs the behavior of all relevant components, services and data in an extended IT environment and simulates them, This capability allows developers to determine how an application based on those services will behave under various circumstances in the real world. Developers can become more agile using Virtualization practices to support increases in complexity, without a corresponding increase in testing and development risks and cost.