• Insufficient understanding

Large organizations are usually aware of this solution, but I wish to refer to a host of small and medium organizations who remain untouched by this technology. I have met and discussed this subject with quite a few IT heads from these companies and find that though they have heard about this subject and understood a little bit, they still do not know enough to recommend and use it in their set-up. They take the safer route of adding a few more servers to take care of extra load and this is easy since it involves an incremental investment. It is best for such firms to retain an advisor who can help the management with a long term technology plan.

• Replacing old servers

The story at some of the larger companies is a bit different. Here we have IT Heads who are better informed and well aware of the subject through interaction with various users and vendors in seminars that they attend. However, they hold themselves back not knowing what to do with existing servers. They say that they need only say two servers and cannot do away with other servers they have and hence cannot invest in something much bigger.

I have often asked them when would a time come when they would have to replace all servers so that they can opt for virtualization; for they will always have servers of different vintage. I could persuade a couple of such organizations to start virtualization with new applications and move the older ones as and when the old servers age and get due for replacement. These organizations now look back and say clearly that they took the right decision.

• Not selecting the right implementation partner

Some CIOs do get carried away pinning confidence in the capabilities of the internal team, thus, trying to do most of the tasks themselves. At other times the selection of the implementation partner is faulty and they appoint one who quotes lesser. This leads to an avoidable problem resulting in a failed implementation or a sub-optimal solution delivery. One needs to realize that the technology is complex and needs to be implemented well to derive clear benefits.

• Extracting full benefits of the technology

Implementation of virtualization requires adequate planning with proper allocation of server resources to various applications in a manner that allows an optimum use of resources and at the same ensuring a good performance for all jobs being run. People are often satisfied with some level of partitioning of the processor pool whereas more can be achieved. The technology also provides a good number of features which lets us extract more out the boxes we possess. For example I know of many an organization who do not use features like cloning, dynamic allocation of processor pool, resource management tools, mirroring of applications enabling fall back in case of failure of a processor, disaster recovery planning etc. A vanilla implementation therefore gives us benefits but only as much and no more.

I will conclude with my opinion that this technology is immensely useful and it is for us to make full use of its features to extract maximum benefits.

Benefits are real; I have experienced them. How much we gain will depend on the mileage that we extract from technology and our propensity to try and innovate. Let me explain various benefits that virtualization can bring in. Following are the key areas where we can see gains:

1. Better utilization of hardware resources:

You can subdivide the processor and assign more than one application to it. For instance, I have often run four different applications running under different OSes, assigning them to a single processor. In the normal course, processors may be fully loaded or partially loaded and therefore we have server resources that lie idle and when we count total idle capacity in various servers, we may find a huge unutilized capacity. Since virtualization creates a pool, we can assign desired resources to each application.

2. Minimum physical space required:

Since we utilize fewer servers the space requirement in the data center comes down drastically if servers are stacked in a rack. In these days when space is at a premium, virtualization comes as a boon to us.

3. Lower power consumption:

Since we use fewer servers, the consumption of power would be low, despite the processors being loaded fully. That reduces the amount of heat generated by servers, and correspondingly, even the power consumed by air conditioning systems is also lower. In one of our recent implementations, we proved that the power consumption could be lowered by 40% with virtualization implementation.

4. Lower investment (capex) on hardware:

Proper capacity planning could see a drastic reduction in the capital expenditure that we incur on server buying. For instance, we had blade servers and by creating a pool across several blades we were able to use the available resources and thus put away the investment proposal that was on my desk. In case we host our applications or use external resources, expenses would come down.

5. Less maintenance overhead of IT:

Virtualization simplifies server management as the entire server farm can be managed through a single console. In addition, recurring expenses on AMC and other costs reduce.

6. Provision of resources on demand:

In a conventional environment, since we dedicate a server for each application, we are at our wit’s end when we have to make available sever resources for any urgent requirement. In a virtualized environment, we can make available additional resources on demand from the pool we create. In additional we can cater to a periodic demand (say month end or year-end processing0 by reducing the resource allocation to any non-critical application and make it available temporarily to the ERP or any mission critical run.

7. Easier to do patch management and upgrade:

We used to create a clone of a server and apply patches on the original file. If for some reason, the new version did not run, we could revert to the old version within minutes. The same process applied to software version upgrades and this was a great facility which made these tasks easier.

8. Disaster recovery planning:

Various features exist through which you can recover fast in case of any interruption due to hardware failure. The entire set of programs can be backed up on a different physical server and can be activated quickly.

How to convince the management

These benefits are ones that I was able to enjoy and once implemented we went about quantifying some of the benefits and demonstrating them to the management. The one question that I am often asked is: How did I justify the investment to the management at the time of taking approval?

I, of course, had problems in working out a cost-benefit; and therefore took a different stand. I worked out figures showing investment that we would otherwise make on additional servers if we did not go in for virtualization: that figure being much higher than the investment in virtualization. It worked and I got the required approvals.

Virtualization comes in various forms viz. server virtualization, storage virtualization, application virtualization or desktop virtualization, but we will talk of only server virtualization here.

What is server virtualization?

In simple terms, virtualization is a method of running multiple independent virtual operating systems on a single physical computer. Let us say it is the masking of server resources from server users, including the number and identity of individual physical servers, processors, and operating systems. All server resources are put together as a pool from which each application or task draws out the resources it needs, thus making the application or the user believe that his task is running on a separate and dedicated server.

How this works is simple to understand. Virtual servers work through a middle layer called a ‘hypervisor’, which masks the servers from the users and takes over the task of allocating resources to each task from the server pool. To understand further let us consider the following statements:

• Virtualization enables us to combine servers from multiple generations into the same virtualized server pool.

• Virtualization allows a group of inefficient servers to be replaced with a fewer number of machines.

• Virtualization is a software that allows a piece of hardware to run multiple operating system images at the same time.

The need for virtualization

In the conventional set up, we keep on adding servers as we introduce more applications in the organization and when the numbers grow very large it becomes unwieldy and difficult to manage. Further, every application consumes resources differently; leading to some servers being strained to the full and some others have it easy with very little usage. This leads to sub-optimal utilization of server resources which is undesirable. Adding to this is the difficulty of provisioning space to host several servers.

How to implement virtualization

If you are thinking of implementing virtualization, I would recommend the following steps:

1. Clearly assess the requirement of server resources in a three-year horizon, taking into account additional applications that are likely to be introduced; and the projected growth of the organization leading to higher number of users and the increasing size of data to be processed.

2. Evaluate and select a virtualization software that is appropriate, given the current IT landscape and size of the set-up.

3. Re-asses the hardware resources and plan new servers that would make use of this new environment.

4. Choose an implementation partner who has adequate experience and expertise in the software selected.

5. Work closely with the partner to work out an optimal design that works best to accommodate your applications.

A breakthrough technology?

I think it is. It is amazing how the researchers worked out a way to break the paradigm of considering the processor as undividable and to create a mechanism by which the processor lends itself for manipulation. It is a wonderful solution to leverage on and I hope most of us use this to make optimum use of the resources.