Virtualisation is here, or so we’re told. But are business’ ready to adopt the technology that allows virtualisation to become a reality? In this article I want to weigh up the immediate advantages and disadvantages associated with virtualising you’re infrastructure and the potential gains and pitfalls moving into the future. I am not going to cover security; this should be at the forefront of any network/storage/server architects mind as it is. Virtualisation is not a new idea. It has been around for many a year, in fact the very first IBM mainframes incorporated elements of virtualisation as we know it today. Another misconception is that virtualisation encompasses server infrastructure alone, this is not true. Many devices from a number of vendors have stand alone virtualisation abilities. For example the Cisco ASA security devices have the ability to create many virtual devices within the one appliance. These are isolated virtual environments, for a better view, a step back is required and a view of the architecture as a whole is in order. It is the holistic view that needs to be taken into account to answer the question posed in the title of this article.

To truly understand the cost, risks and advantages attributed to virtualising your infrastructure I am going to break the infrastructure down into four parts: Application, Servers, Storage and Network. Older organisations have had IT thrust upon them in the last 15 years, the infrastructure, unless well managed, will have grown organically with multiple vendors having sold a variety of technologies and platforms into the business. It depends where your infrastructure currently rates on the ‘organic sprawl’ scale as to the cost to bring your physical infrastructure to a point that would allow for virtualisation to become a reality.

Let’s take a look at some of the high level advantages of virtualisation. By simplifying the physical architecture and bringing the complexity into the software, the immediate advantages start to become clear. Fewer devices = less power, cooling and space. Pretty standard stuff if you talk to anyone au fait with virtualisation techonlogies. Rather than focus on short term cost reduction, what would you think if you were told that with the ability to pull available virtual resources, anytime from any area of the infrastructure, you could feasibly build blueprints that can be used over and over again? Consider there’d be no time required  for bare metal boxes to arrive, no time needed to install applications, no need to reconfigure or expand the network and no direct requirements to build new LUN’s for applications.

So where are the disadvantages? To start, the initial capital expenditure will be high, even for the most savvy of business’ that have standardised on certain vendors, the move from a siloed infrastructure to a logically siloed infrastructure is costly. The servers need to be of a certain spec, usually with far more processing power and memory, therefore higher cost / server along with the cost of disposing of the old hardware / writing it off. By bringing multiple virtual machines onto one blade or server; the network connectivity will need to be closer to 10 GB than 1 GB. This could possibly involve ripping out the current network and re-building the core from scratch. Looking at storage, Unified Fabric switches and Fibre Channel over Ethernet (FCoE) are now worming their way into the market place. Utilising these technologies as well as some inherant capabilities that certain vendors offer can drive the cost/mb down significantly as well as allowing dynamic storage to be placed alongside networking and server components in the virtualisation blueprint.

Let’s say that you have gone through the above, you have a brand new architecture, what next? Attention now needs to be on the business itself, by removing siloed infrastructure, a fundamental change in business process is required to fully take advantage of the potential return on investment laid on the table. The agility the new infrastructure offers requires agile thinking and leadership to accompany it. The business will need to go through a huge change in mindset.  Acquisitions for example; with a blueprint in place it is entirely possible that the incorporation of the acquired business could take less than 3 weeks! The technology is there to facilitate it, but are the people?

Despite the advantages, a badly managed virtualisation project can be disastrous. If you thought that server sprawl was a bad thing previously, imagine an unchecked IT team creating hundreds of virtual machines in a fraction of the time it would have taken them to fully configure a bare metal box previously. Bringing the complexity into the software presents a unique set of challenges in itself, 10 GB core or not, saturation will lead to organic sprawl once more, just this time with a price tag to weaken any mans knees.

A clear, defined strategy from day one covering all aspects of the business, technology and people is the only way to guarantee the highest possible return. Ready?