A detailed study of the existing network infrastructure was undertaken to understand the call volumes and patterns. The IT team collated data on telephony usage and the growth pattern across various offices over a period of six months.
"To enable people to communicate in an optimal manner, we decided to introduce enterprise-wide VoIP and invest in large telephony infrastructure," says Hariharan.
The team designed the VoIP architecture, ran proof- of-concepts and tested various call volume tools. The detailed study and the trend analyses helped the IT team in designing PBX migration roadmap, VoIP trunks and bandwidth requirements.
To cut costs on migration of PBXs to the newer VoIP ecosystem, the existing PBXs were VoIP-enabled by software upgrades. Using logical partitioning (a subset of computer hardware resources, virtualised as a separate computer) the existing fleet of PBXs was leveraged to route both VoIP and PSTN calls across the enterprise.
The revamped infrastructure went live in June 2007. Hariharan and his team proactively accounted various roadblocks that could have possibly cropped up during the entire phased deployment cycle. Redundancy was one aspect that stood out like a sore thumb. "Since we wanted to accomplish our goal of rolling out the new VoIP-enabled network without any downtime, we stressed on building a resilient network with adequate redundancy," says Hariharan.
In addition to the three carriers being roped in to check redundancy, connectivity between US and India were triple-routed. With all his systems in place, Hariharan also had to connect continents and cross oceans. While submarine cable connects were chosen over the Pacific Ocean and the Atlantic Ocean, a third route was devised over satellite. The locations in India networked to US through a direct route, another routing through Singapore and the third going through Europe. "This way we removed any chance of harboring a single point of failure and minimizing breakdown instances," he says.
The IT team had expected certain performance issues with the new VoIP ecosystem. To figure out the threshold points, before roping in senior management, the systems were rolled out to a facilities and administration team of about 400 users. This eliminated any practical or perceived problems with the systems. "We had suspected some performance issues with users at remote locations. With the feedback from the pilot users, we resolved performance issues by leveraging quality of service and by upgrading the bandwidth marginally wherever required," says Hariharan.
On the IT infrastructure side, virtualised environment riding on VMware tools was brought in. A number of underutilized servers across three distinct datacenters in Mumbai were collapsed into one box sitting at the new, consolidated datacentre. Once the projects were over, the virtualized setup was reused and the resources were allocated to other projects in the pipeline. Currently, VMware supports a total of five applications on 27 virtual servers. This has naturally led to substantial savings in terms of procurement of systems, real estate space, and reduced power consumption.
Sign up for Computerworld eNewsletters.