Driving on the Cloud

Fleetwood, a recreational vehicle manufacturer approached Allied Digital to get their data center managed. The data center was burdened with redundant applications, certain components were running thrice and all of this took up 170 servers. The optimization process required Allied Digital to manage the landscape of several backend software applications.

The deployment

After surveying the damage, Allied Digital, led by Sanjiv Patki, Global Vice President — Managed Services observed that the backend was on Oracle; there were two instances of middleware-Websphere and Weblogic; almost all apps ran on Weblogic but the employees had a soft spot for Websphere. Allied Digital had to point the new database to Oracle-based Weblogic removing the outdated version of Websphere that was using mission critical resources. Allied Digital's team started to shut off components that weren't in use and cut down the number of servers from 170 to 119. Allied Digital then discussed the private cloud with Fleetwood, convincing them of a minimum of a 20 percent drop in resources.

Shifting to the cloud

“We distributed the company into three sections: the IT staff, the IT managers and CTO, and finally the top management,” Patki said. Patki's team then took stock of the data center once again. Each application was reviewed and each deployment remapped. Allied Digital offered Fleetwood two options: one to use a private cloud, while making a substantial investment, and second the often-used pay-per-use model. Fleetwood chose the recommended private infrastructure arrangement. After three months of discovery, Allied Digital managed to eliminate processes bringing down the server number even further by automating several processes.

Challenges

Allied digital had been working on a model to implement but had been unable to find a way around the elaborate 70x70 matrix that stood before them. “While we were designing the technology to use, we faced vendor challenges,” Patki said. His team listed vendors but couldn't find a way to focus on one. Patki realised the time spent was increasing exponentially due to the reluctance of the Fleetwood staff to align with Allied Digital. After the initial three job cuts, Fleetwood were hostile towards Allied Digital while, rightly predicting further job cuts. “The Fleetwood IT staff were cautious about this, while we identified some of the staff as customer specific and the rest that were being left redundant were absorbed into Allied Digital,” Patki said.

The other hiccup faced was the major issue with Microsoft software that comes up when clubbing applications. Fleetwood then had a read-only database coupled with a transactional database. The e-commerce applications ran parallel to payroll applications on one physical server. Patki decided the make-up of several of these servers had to change and a data warehousing structure had to be established. Since Fleetwood was an organization that allowed consumers to book their portable mobile homes online, any downtime would mean loss in revenue. “We decided to club the server handling transactions to process the batch specific functions at night,” Patki explained. Since transactions were mission critical process and needed high computing power, they were assigned a physical server, but regular users were dumped on a virtual server, spread across two physical machines.

Cloud benefits

The biggest saving for Fleetwood was not the green IT initiative or the recovered costs, it was increased performance.

The online transaction response improved by 15-20% along with the batch processing time that decreased by close to 20%. The redesigned data center, which had 119 servers, now had just 18 blade servers, two sandboxes and a tape library. “An infrastructure of about 30-35 racks dropped to three,” Patki said. There was an automatic increase in real estate that could be utilised for other purposes. With the decrease in servers, the cooling costs tumbled by 30% and the power bill by 40-45%. By routing the entire operations to an offsite virtual server for one night, the backend rehaul could be made easier and the downtime was reduced to zero. “This entire process demonstrated a 50-55% savings in IT costs,” Patki concluded.