We turned to Focus to get some perspectives on this topic. The views are diverse and should give some insights into this ever-evolving discussion about managing business critical apps in a virtualized environment.
A common theme is the difference between physical servers and a virtualized environment. The difference may seem basic to some but the difference points to what can come when applications are no longer tied to a physical servers.
The challenges are many, but the most important one is that unlike physical servers you cannot infer the performance of applications running on virtualized servers by looking at resource utilization statistics like CPU usage, memory usage, and I/O (network and storage) rates.
You need to assess the performance of your infrastructure by using tools that tell you infrastructure response time (end to end latency for all requests) and applications response time. This means you need to throw out all of your legacy tools and start over.
David Lynch, the founder of the FIOS Group states that virtualized environments require a combination of change management, release management, configuration management and optimization.
In his view, integrated systems can be more effective than a collection of tools. The programmatic nature of virtualized environments means that it can be difficult to use multiple tools when there are issues such as capacity, performance and sprawl.
Jonathan Eunice, principal IT advisor with Illuminata, maintains that you need to manage the applications, not just the virtual environment:
Running them virtual improves some of their characteristics, such as infrastructure utilization, flexibility of deployment, workload mobility, and so on. But it doesn't change the fact that you still need to monitor the apps and services themselves. Their throughput, responsiveness, availability, and other operational characteristics are just as important as if they were running on physical infrastructure.
Indeed, monitoring app/service characteristics is probably more important once you go virtual, given the relative newness of virtual infrastructure, and the typical need to demonstrate that the apps are running just fine on a shared infrastructure. You can only do that by monitoring app/service characteristics and reporting them in business-relevant terms.
This is an evolving discussion. The enterprise is more interconnected than it has ever been before. Apps are migrating to virtual networks. But that's just the first step. The bulk of the work is in managing the virtual network and the apps that sit on top.