Transitioning to a Virtualized Data Center

These days, it’s fair to say that just about everyone gets the value of virtualization (and, for those that are still on the fence, its potential value).  Organizations can differ dramatically in their implementation of virtualization.  When implemented correctly, virtualization helps reduce complexity, reduce management overhead, and increase overall operational efficiency.  Those words should be music to the ears of any data center manager that’s struggling to make their business requirements and technology ends meet.

While server virtualization tends to grab the majority of headlines and mindshare, the general approach of virtualization – adding layers of abstraction – can apply to many different core data center areas.  Those areas include processing, memory, storage, and networking.  Common methods of virtualization include the use of network-layer abstraction that encompasses both physical machines and VMs, storage-focused features, and a dynamic and highly automated core infrastructure.

So, what’s the best way to move towards virtualization?  There are several approaches that organizations can take.  One approach is to “rip and replace” – to retire existing infrastructure components with new ones that rely on a virtualized infrastructure.  This can be a costly approach, but if the old servers, storage, and networking gear are overdue for an upgrade already, it might be a valid approach.

More commonly, organizations will want to take smaller steps towards virtualization.  Options include upgrading key, critical systems, or deploying support for new applications and workloads on new infrastructure.  Let’s look at some specific ways in which IT pros can move toward greater levels of virtualization:

1)      Invest in virtualization expertise:  A common barrier to virtualization adoption is a lack of knowledge and experience of working in virtual environments.  Common virtualization platforms make it easy to get up and running, but without some depth in understanding different types of virtual disks, methods for migrating storage and VMs, and issues related to snapshots, organizations can end up with many new problems.  In some cases, moving to a virtualized environment without the necessary expertise can be worse than sticking with the ‘tried and true’.  I have seen this in many environments where performance tends to suffer due to sub-optimal configurations.


2)      Move workloads to VMs: Among the many benefits of having your workloads running within VMs is the ability to move them between host servers, as necessary.  By encapsulating an entire OS and all of its applications, you can more quickly scale its CPU and memory requirements, and make adjustments when needed.


3)      Invest in centralized storage:  The need for storage is never-ending in most organizations.  Rather than adding smaller, independent storage arrays (based on locally-attached storage), you can gain efficiency by investing in storage arrays.  The initial hardware investment can be costly, so it’s important to identify your storage requirements from a cost vs. capacity vs. performance perspective.  Not all of your applications will require the highest levels of throughput and fault-tolerance, so invest in more cost-effective storage features wherever possible.


4)      Invest in high-availability:  By storing your virtualized workloads on a compatible storage infrastructure, you can use a consistent method of achieving automatic fail-over for high-availability.  Contrast this approach to one in which you use a different method of H-A for your database servers, messaging servers, and application servers, and the simplification can significantly reduce complexity.


5)      Support self-service: Some of the most mundane tasks that are presented to systems administrators – for example, provisioning new servers or making minor configuration changes – can also be the most important.  If the infrastructure doesn’t meet users’ needs, it can result in a significant loss in productivity.  To offload some of this work, data center administrators can deploy self-service web sites that can allow users to provision their own virtual machines, as needed.  Of course, this shouldn’t be an all-you-can-eat VM buffet.  Administrators must define resource usage quotas, create IT-approved “gold” master images, and help ensure that users are allocating their resources wisely.

These are just a few examples of ways in which you can “phase in” virtualization-related features.  I focused on those areas that are likely to provide the quickest return on investment, but this is only scratching the surface.

Regardless of your approach, though, it’s important to keep the primary goals in mind: By leveraging virtualization technology, you can simplify administration, increase hardware utilization, and improve IT agility.  Ideally, these investments will pay for themselves quickly.

Tags: Data Center,Technology,Virtualization