Configure with Extreme Prejudice: How to Ensure Virtualization ROI through Enhanced Management

By Stephen Beaver
published: Friday, July 17 2009

Today there are several driving forces creating new opportunities for enhanced management and automation in virtual data centers. The vital need to increase internal responsiveness, meet rigorous compliance requirements and ensure virtualization continues to deliver on its cost-cutting promise is driving IT to look at configuration management and automation processes and tools for their virtual data centers.

The Face of a VI Admin/Engineer

VI engineers or administrators have typically come from a Windows environment and are pushing virtualization. They like to create new virtual environments that solve business issues and to be the hero, yet they aren't necessarily interested in managing these environments. For purposes of this article, we will call virtualization engineers and admins, the average Joe. Joe has been riding the virtualization wave since approximately 2006, consolidating servers and reducing costs for his company. In fact, he's flaunted his ability to save his organization hundreds to millions of dollars in server and licensing costs by consolidating servers with virtualization. What Joe isn't telling management is that he has no idea how the 5,000 VMs in his data center are configured or why several went down last week.

The Facts and What Studies Show

According to a recent Forrester survey of 164 virtualization professionals in medium to large sized organizations, many see lack of virtualization management tools as a barrier to more extensive virtualization deployments. 64% of respondents cited that they'd like to run more VMs per server. 57% of respondents expressed significant concern that application failures in one VM could affect other VMs in the future and that their top issues with managing virtual environments included being able to prevent critical events from affecting performance or availability. Despite IT's efforts, problems inevitably occur - which is why respondents also cited that diagnosing problems ranked close in importance to preventing them in the first place. Bottom line, Joe doesn't want to have fingers pointed at him for issues that arise in the virtualized data center.

How Does Virtualization Impact Configuration Management

Indeed, virtualization has delivered generous benefits to its users, but compared to its physical counterpart, virtualization does introduce a new dynamic that if not managed properly, can open up Pandora's Box. The topic of VM sprawl has been covered and IT is aware of the issues surrounding it. But what hasn't been discussed in great detail, is the life span of a VM which can range from minutes to years, and the importance of configuration management and the ability to troubleshoot problems from conception to death. A lot of organizations have used virtualized physical servers that are coming to the end of the physical hardware life cycle but have not given any thought to the end of life of a virtual machine.  So how is virtualization impacting configuration management? The speed and scale for which virtualization is renown is generating a new level of complexity and impacting configuration management more than ever. The ability to take things down and put them back up and move them around makes for an extremely dynamic environment. IT is moving faster than it has in the last 10 to 20 years and is now beginning to panic about VMs being able to increase the efficiency of their virtual environment. Knowing the relationship between the various components in an IT infrastructure is paramount, and new emphasis is being placed on ensuring accurate provisioning, patching and application release controls.  Why is this the case? A single instance of a VM can have thousands of configurations with many dependencies. If load balancing isn't optimized it can cause performance degradation. If the DRS isn't configured properly, major performance issues can arise. In the past, trying to find out how many VMs have multiple CPUs associated with them required manually looking at the properties and writing them down on paper. With more than 100 virtual machines, this isn't very efficient.

In the physical world, IT puts an agent on a machine and people knew where the machine resided in the network. Virtual machines however, don't have the same level of rigor and standards around operating systems, patch management, updates etc.

The questions that are now running through IT's mind are what machines are running, why are they running, what is their purpose, how fast are they changing, what threats might they pose and who created them? The distinctive differences between physical and virtual machines are their containers, formats and the mechanisms for moving them in and out of their states.  If they are unable to have visibility into these issues, their SLAs and the credibility of their virtualization projects are in jeopardy.

Configuration management is quickly becoming one of the most important disciplines for IT because it not only answers the questions IT is grappling with, but if the right tools are in place, configuration management can actually proactively mitigate security, compliance and availability risks. Some would argue that if their machines aren't in a production environment, why does high availability and release control matter?

Downtime in the Real World

I had the opportunity to meet with a sales engineer at a major technology company recently. He disclosed an interesting story that helped me understand how a small issue with a VM can impact a major organization. He went to check out a demo stored on a VM so that he could assist his sales rep in closing a deal. He proceeded to his network path to check out the demo, but the VM was down. Not only was he panicked and worried about not being able to show a potential customer a demo, he placed several calls to IT to understand when the VM would be back up, and several calls to engineering to help him build a new demo as fast as possible. When we discussed the amount of time and resources required for what should've been a simple task, I was shocked. It is clear that at any level in an organization, downtime of a VM can have a significant impact. While it clearly isn't the Exchange server that went down or the order management system, it did impact resources and time significantly.