A 'virtual' peek into Bajaj Auto's dashboard

As its business sprawls across Indian and Indonesian factories, the company operates three datacenters, that includes a primary one and a co-located server site besides a DR site. By May 2009, the company was seen running most of its applications from an infrastructure at its primary datacenter comprising 30 virtual machines.

Raghuvir Singh Sohal, manager, MIS, for Baja Auto Ltd, shares his experiences and views on pilots, virtualization, AMCs etc, in a brief chat with CyberMedia News

You did a proper pilot before deploying virtualization at your organization. How well did it work?

It was a very useful mode. Because when we first decided on virtualization, we were slightly hesitant. With the initial try-out, we came to know what hardware resources we are getting into and we could also gauge the real performance impact taking into account the heavy load and dedicated RAM factors.

Did it pay off?

Yes, the number of racks required has come down from four to one. It has also resulted in a cut in server provisioning time from two months to a day now. It has also helped improved server hardware availability to 99 per cent and above, and enabled centralized management of datacenter infrastructure etc.

It has freed up old hardware resources and helped especially on AMC related machines, their hardware gestation issues etc that were in out-of-warranty mode. It’s also easier to buy a new hardware with a cost calculation idea across three to five years. Earlier dedicated hardware resources for each application were required.

Any flip side besides the benefits you gained?

We have had some old (Novell) systems. Some of them were virtualized effortlessly, while in some cases, it couldn’t happen, possibly because of older environment and OS issues. Each software has its own unique constraints.

Now that you have tried virtualization at your data centers in Pune, would you consider that option for your ERP disaster recovery facility in Bangalore?

We haven’t gone in that direction yet. It would be too premature to talk about that now. We won’t be interested or experimenting in that area at least for now.

How do you decide on a technology choice?

My focus is on management of information systems and I always emphasize on whether a real business need is being satisfied or not.