We think virtualisation is to IT what digital money is to commerce. Its simulated computing environment that abstracts away from the physical hardware offers great flexibility in provisioning computing resources. To take advantage of cost arbitrage in an area where competition is only based on cost arbitrage, no organisation can ignore efficiency.

IT enterprises worldwide agree upon the need to design a lean infrastructure, providing more for less. It is also time to look at infrastructure as an adaptive system to address upcoming challenges. Virtualisation is at the heart of building a lean infrastructure and therefore is almost unquestionably popular. Cybage has its own ExcelShoreTM system that helps us in providing better value-per-unit cost. This project is a component of the ExcelShoreTM.

What?
The project was to optimise infrastructure or have a lean-infrastructure. Usually, this lean infrastructure is aimed towards server farm and that’s there in place for many organisations in some or other form, including Cybage. However, here this virtualisation application was to use server based computing or virtual desktop to build a lean client/desktop infrastructure to reduce cost by optimising desktop setup. The remote desktops mirror the applications, programs, and data on their respective virtualised desktops on the central server. Thus computational needs of various users can be accomplished using a thin client comprising a monitor, keyboard and mouse. Direct saving comes from reduced machine requirements and space (and related utilities and facilities requirements), but there is fair amount of savings through by-products such as simplifed provisioning of resources and reducing downtime.

Why?
We have a small, but diverse Customer/Technical Support and Back Office Processing group of about 250 people. The group works 365 x 24 x 7 but not every customer needs that schedule. There are customers who need support only during weekends but 24 x 2. There are customers who need support during all days but only 16 hours (16 x 5) and so on. Considering the diversity of schedule, it’s not easy to optimise use of infrastructure. Ideally, optimisation in Customer/Technical Support or BPO is done by sharing the same machine and cubical by different agents during different shifts. In Cybage’s case, this is difficult due to various disparate shifts and small team sizes. The need was to optimise space and machine usages while keeping one machine per cubical at the most for different shift users.

How?
We started optimisation efforts first without virtualisation and soon we realised we cannot optimise it much due to very different schedule requirements and small team sizes. For almost 250 people, we could save only a total of 6 cubicles which was not much. This started on the request of our COO, and the amount of time that we spent and result that we got forced us to think beyond usual the workflow/roster initiatives.

Beginning
Sometimes, a dead-end presents a better path, and this is so very true in this project. I remember, when our COO said, “This is a technology problem and you guys solve it. We are not operating with desired efficiency of three shifts per cubical and we need to achieve that.” That was the beginning for us to find a solution which works in given situation and not necessarily in an ideal situation of 365 x 24 x 7.

Ideas started flowing in and included running a Citrix-based environment to a file-server-based approach to a traditional Novell-based setup of client-server to laptop versus desktop – all had some or the other issue and were not resulting in the desired output.

Breakthrough
Cybage was building its expertise around Cloud computing at the same time. Closely knit emerging technologies team at Cybage helps in exposing and validating thought process across project teams. This was the time when quick provisioning and mobility, functionality sharing, central hardware and software management, green computing advantages of virtualisation led our project to desktop virtualisation.

Quick provisioning and mobility: Virtualisation is most suited to our case where we have employees working in shifts and we don’t want to bind them to a particular cubical. Operational optimisation of the existing hardware and provisioning of computing resources can be easily achieved through virtualisation. The remoteness of the data centre ensures that data security is not compromised.    

Functionality sharing: The disproportionate use of thick clients in that the functionality of the thick client is not fully exploited can be overcome. The user hardware requirements can be provided on need-basis using software. Along with hardware, software licensing costs are also significantly reduced.

Central hardware and software management: A looming infrastructure is definitely an overhead for organisations. Virtualisation can automate the process of managing the hardware and software requirements centrally.

Green computing: Normally, a thin client uses only approximately 20 per cent of the processing power of the CPU – this increases power consumption. With desktop virtualisation, the processing power of the CPU is effectively utilised as all the processing is done at the server end. Thus overall power required for running and cooling the hardware can be minimised, also reducing the carbon footprint. This was not the main success criteria for the current project, but it has now become a full-blown initiative.

Constraints
In the hyper-competitive IT sector, a business enterprise needs to make a calculated decision before adopting a new technology. It is really unfortunate when a company pumps huge investments into procuring a technology that then fails to deliver sustained business value. The two key concerns that need to be kept at the top for any optimisation projects are utilisation of the existing setup and ease of deployment and use.

We looked at many virtual desktop infrastructure (VDI) solutions but we realised they mainly focused on server side technologies and used either a thin client or a thick client per user to connect to the data centre. A thin client is a feasible option for a startup organisation, but for a large organisation that has already invested in procuring hardware it is impractical. A thick client per user is a waste of resources when you want to optimise machine usages.

We adopted a very unique approach by which we convert single thick client into multiple thin clients by sharing its hardware among multiple users. Each user in Cybage VDI solution has only a monitor, keyboard and mouse. And, there is single CPU shared among N users. The CPU or the server hosts discrete user sessions and allows users to access their virtualised desktops on the data centre.

The deployment of our VDI solution does not involve any kernel level changes on the client side. It seamlessly fits into the existing infrastructure without adding to any maintenance overheads.

This project was not a volume project. It was a high-complexity project coupled with practicality to extract the best cost benefit. On the technical side, it was led by one of our two senior-most architects to ensure the solution was technically the best possible one given the constraints.

Achievements and the Future
The project was successfully completed, and we have started realising its benefit in term of cost. The direct saving here is the optimum usage of machines and cubicles (with related utilities and facilities), which translates to approximately 8-10 percent of revenue for given number of cubicles. There is no Capex saved in our case as there is a buffer infrastructure created for overall growth. However, if we look at the next three years, it does reduce buffer infrastructure requirement.

Being a private company with a very active management, Cybage is able to focus on long-term issues and that does help in executing projects which help in the long run. VDI was even easier to get buy-in for, as this had a visible and quick benefit.

By Jagat Singh