Transforming to a Lean Data Center

By Prabhaker Yasa, Vice President-IT, JDA Software

Data centers are the backbone of any organ­ization’s IT infrastruc­ture. Managing data centers, the peripheral environments and re­sources are a part of the entire data center management activity. The as­sociated costs typically account for a large part of an organization’s IT related spend.

Over time, data centers develop efficiency issues like poor hardware resource utilization, wasted energy and obsolescence of equipment. By transforming into Lean Data Cent­ers, organizations can gain efficien­cies through lower capital and oper­ating costs.

"Consolidating legacy servers has the added advantage of lowering maintenance costs while also providing an opportunity to bring in higher end configurations"

Lean Data Center transforma­tion involves understanding the drivers, methods and tools that op­timize the footprint of data centers while reducing engineering com­plexity and the associated support required. ‘Lean’, as a philosophy, focuses on distinguishing between value added and non-value added activities with the aim of eliminat­ing waste. Waste can be in the form of defects, overproduction, wasteful transportation, waiting time, excess inventory, unnecessary movements and over-processing.

Adopting a lean philosophy for data centers can result in reduced physical inventory, streamlined pro­cesses, improved technology land­ scape, better space utilization and reduced power consumption, trans­lating to lower capital and opera­tional costs to the business.

During JDA’s lean data center transformation, the technology teams adopted the lean six sigma framework and tools to derive. The need to consolidate physical and virtual servers was felt based on the server-to employee ratio. Sponsor­ship was obtained through cost-ben­efit analysis of the potential savings from lean initiatives against the To­tal Cost of Ownership (TCO) of our data centers.

A portfolio of lean initiatives of varying magnitude of scale was put in place to achieve the desired bene­fits. This helped identify focus areas for lean and continuous improve­ments, in a structured manner. This also addressed the necessary objec­tives, resources and risks that would come as part of the transformation. All sub-teams committed the nec­essary resources to the program and trained on lean philosophy and tools, to gain momentum.

An integrated change manage­ment plan was developed to ensure that all stakeholders have a buy-in as well as visibility on the various ob­jectives set including any post-transi­tion issues like service response, case resolution and uptime SLA adher­ence.

Consolidating legacy servers has the added advantage of lower­ing maintenance costs while also providing an opportunity to bring in higher end configurations. This also helps deal with vendor issues like end-of-life support, security and break-fix patches.

Extensive data was gathered on server hardware utilization and soft­ware usage over a period of time. Pareto analysis, Kanban and visual management techniques helped identify focus areas. Identifying and categorizing the “vital few” versus “useful many”, ensured clear prior­itization of the workloads based on actual usage of the physical and vir­tual servers. Only 20% of the servers were of the ‘vital few’ category and were being used for critical work­loads, providing a huge opportunity for consolidation of the rest.Using partitioned and virtual server envi­ronments we could simultaneously reuse unused capacity and process­ing power, thereby eliminating pro­liferation.

As the organization was increas­ingly moving towards cloud based services, identifying the non-critical workloads became pertinent. This required that workloads were al­ready streamlined and optimized prior to the move. Our co-located data center optimization exercise en­sured that we could achieve a good balance between our physical and cloud footprint.

Once the infrastructure was con­solidated, the focus for optimiza­tion shifted to operational processes. Value stream maps (VSM) was cre­ated to document the current state, non-value adding activities, value adding activities and resulting waste in the processes. The desired future state was then arrived at with active involvement of the business users. Bottlenecks and pain points were identified, providing a clear visibility of the series of activities that deliver services from the IT Service Catalog to the business users. (VSM) were created for the server resource pro­visioning process and self-provision­ing was identified as a key step for process simplification. Delays due to case hand-offs and unwarranted ap­provals were reviewed and targeted for reduction.

As a final step, FMEA (Failure Mode and Effects Analysis) was per­formed to identify potential single point of failures like backup, fail over, HA (high-availability) and other risks. This ensured that there was no degradation of service upti­mes, resilience or service levels for the business units.

Results

As an outcome of the program, the server-to-employee ratio was re­duced from 2.5:1 to 1:1. This also helped setup a technology frame­work for further optimization in the future. Many terabytes of SAN storage, thousands of processors and terabytes of memory were released. Ourlean data centers now provide us with higher utilization, improved compliance, streamlined provision­ing process and lower optimal oper­ating costs.

Don't Miss ( 1-5 of 25 )