In Part Two of BSMdigest's exclusive interview, VMware's Javier Soltero, founder of Hyperic, talks about virtualization and BSM.

BSM: How is virtualization changing performance management? What are the main challenges that didn’t exist before?

JS: The same thing that makes it extremely convenient to rely on virtualization - the ability to treat what used to be a full piece of hardware as one large set of files, and move them around - that introduces a significant set of new problems into the data center that need to be solved.

The first problem is the potential for sprawl, which I think has been amply debated. As soon as you remove the physical obstacle from the provisioning equation, as soon as you make it easy for people to spin up what used to be complete physical systems, as soon as you make it as easy as clicking a button on VMware Lab Manager, you have greatly accelerated the pace at which new servers are going to be put into service.

The second problem is that, even though you have made it easy to stand up what used to be full-on hardware, what about the applications running inside that hardware? Previously, a lot of management tooling for virtualization was simply concerned about getting the hypervisor’s perspective on the health of a virtual machine, which amounts to: is the virtual machine in a healthy state, and how much resources from the hypervisor is it consuming? That is necessary but not sufficient. So that is where the second problem begins. You are running an Oracle database, or some other sophisticated software component inside of a virtual machine. The behavior that software component is displaying is intricately related to the fact that it has an operating system that’s running on top of a virtualization layer. You have to be able to put all the sides of the picture together to determine if it is running correctly, and what to do about it when it is not.

This is the black box effect. People think virtualization means they don’t have to worry about the health of the internals of any of these VMs because they can just restart it, if it dies. Or they can just clone it if they need a few more. That is not necessarily going to help you solve the problem. It may make things worse.

BSM: How do you solve the problem?

JS: The first and most important point is to provide visibility into the guest operating system and into its applications, with as little impact as possible. You need to be able to see what’s inside that virtual machine and then correlate it back to what’s happening on the hypervisor level.

The second point is to use that level of visibility – from the metal through the hypervisor through the guest operating systems and up into the application and its custom componentry – to drive different models for managing your infrastructure, and to make better decisions about the scalability of your software.

For example, if I have a pool of Tomcat servers running on different VMs and I want to be able to shrink that pool when the application is not that busy, I can use the metric for busyness of the application - Tomcat’s request per minute, for example – as the basis for telling the hypervisor to reduce the number of instances, or alternatively, add more. That is a new innovation. That is something you could not do in an environment like this if you did not have this type of tooling in place.

BSM: It seems with cloud computing that the user loses even more visibility. How does the user maintain control of performance management in the cloud? Or do they simply relinquish the responsibility to the cloud provider?

JS: That is a great question and one that we have spent a lot of time trying to answer. I don’t think that debate has been settled but I can give you my opinion. The idea that as the owner of an application you are simply relinquishing all responsibility for that app the moment that you upload it into Google App Engine or Amazon – to use two public cloud examples – and that everything is going to work just fine because these are big name brand service providers, that is hogwash. That is just not going to work, and I think people know that at this point.

The appeal of these cloud services starts from the fact that the end-user of the cloud is now empowered to do almost every part of the application lifecycle unassisted. So a guy with a credit card and a dream can stand up a multi-tier application on a cloud environment without any assistance from operations, presumably. But as soon as that application becomes critical, and has its first problem, that user who previously had no direct operations responsibility will be stuck having to play the operations role. People refer to that emerging role as DevOps. So there you have yet another shift in IT management that requires a whole different set of tools and processes - and of course we are looking at that problem quite intensely at VMware.

BSM: Let’s talk more about the future. How do you imagine virtualization management evolving?

JS: I certainly expect there to be continued diversification in the classes of management tools that exist out there. Spring Insight is the best example of what I mean as far as different tools. That actually has little to do with virtualization and more to do with the fact that as soon as you successfully implement virtualization and our vision of developer productivity - in other words the self-service developer can stand up an application on his own and get it going - you are going to need to have more tools the developer can use to truly understand how that app behaves at scale, before rolling out the app. And that’s what Spring Insight does. It helps you as a developer, while you are building your app, to give you very rich but developer oriented operational perspective on what’s happening inside your app, with the hope that you will use that information to fix problems before you deploy the app to production.

Click here to read Part 1 of the interview with Javier Soltero.

About Javier Soltero

Javier Soltero is Chief Technology Officer for Management Products for SpringSource, a division of VMware(NYSE: VMW). SpringSource is the leader in Java application infrastructure and management, and provides a complete suite of software products that accelerate the entire build, run, manage enterprise Java application lifecycle.

Previously, Javier was the co-founder and CEO of Hyperic, which SpringSource acquired in May 2009. Prior to co-founding Hyperic, he was chief architect at Covalent Technologies, where he led the design and implementation of multiple enterprise products, including the configuration management product for Apache and the Covalent Application Manager (CAM) — now Hyperic HQ.

Prior to Covalent, Soltero was a senior software engineer at Backflip, where he met Hyperic co-founders Charles Lee and Doug MacEachern. Soltero also held senior engineering positions at Netscape, where he participated in the design development of e-commerce and Internet infrastructure suites.

Soltero is originally from San Juan, Puerto Rico and received his BS in Information Systems and Industrial Management from Carnegie Mellon University.