How to Use Virtualization to Save Money and Improve Service Quality
How to Use Virtualization to Save Money and Improve Service Quality
Our IT organization began its virtualization and automation journey with a limited number of key objectives. We wanted to obtain the maximum return on existing investments in server and storage resources and become more responsive to the needs of our software developers. We were pleasantly surprised to discover significant side benefits that far exceeded the objectives of our original plan. These side benefits included labor savings within the IT organization and productivity gains within our R&D teams.
Our IT organization achieved the typical virtues of virtualization such as improving the utilization of existing assets and minimizing additional hardware purchases. For example, server reuse by multiple development teams saved more than $5M in avoided procurement costs. These efforts also reduced power consumption and data center floor space requirements by 20 percent. The financial payback of our investment in virtualization, which was approximately $2M, was achieved in less than six months.
Automation procedures that were implemented as part of our virtualization initiative dramatically improved our ability to respond to developer requests for data center resources. The speed and predictability of our response times, in turn, has made the developers less possessive of "their" assets and more willing to return underutilized resources to the virtual pool for reuse by others. Developers have become increasingly confident in our ability to satisfy their needs on a just-in-time (JIT) basis.
Although it's difficult to precisely estimate the time savings for developers through the new provisioning process, lead times for individual requests have been reduced by at least 10 to 15 workdays. This is equivalent to a collective reduction in lead times of 360 work months per quarter for a 1,500-person development organization operating in five distinct locations.
Launching Our Virtualization Initiative
Launching our virtualization initiative
Our four original objectives were to achieve higher overall levels of server and storage utilization, minimize additional equipment purchases, reduce server sprawl and improve developer productivity. The initiative employed state-of-the-art service request management, hardware virtualization, and system monitoring technologies to manage server and storage assets on a pooled basis.
This approach created an in-house, on-demand, cloud computing environment for our R&D team and provided them with JIT access to the data center resources required to conduct product development projects.
We implemented IT Infrastructure Library Version 3 (ITIL V3) processes for capacity management, availability management, service level management and service reporting to obtain the maximum business benefits from the technologies deployed. These integrated management processes were linked to our hardware procurement procedures to ensure that excess server and storage capacity is available to satisfy developer requests.
For individual projects, developers typically require eight to 10 discrete environments that consist of multiple servers and storage volumes. Without this provisioning capability, more than 100 development projects being conducted concurrently on a worldwide basis could be adversely impacted by delays in server and storage availability.
Achieving Success with Business Service Management
Achieving success with business service management
Business service management (BSM) is a comprehensive approach and unified platform for managing IT service delivery. At the very beginning of our virtualization initiative, we established a comprehensive BSM framework for managing asset provisioning. We employed the ITIL processes and a collection of BMC, VMware and EMC software products to manage server and storage assets in a closed-loop fashion.
Developers now consult a catalog of standard environments and submit asset requests through a standard service request management (SRM) procedure. Requests are routed to IT support teams in multiple geographies and a combination of BMC, VMware, and EMC tools are employed to allocate the requested resources. BMC tools are subsequently employed to monitor utilization of the provisioned assets. Assets are returned to the pool at the end of an individual project or if utilization is observed to be low or intermittent.
We have been able to virtualize roughly half of our global R&D server environment, reducing the number of physical servers we manage by roughly 50 percent over a two-year period and replacing the retired physical devices with virtual machines. One of the hidden benefits of the initiative has been the elimination of older data center equipment that was difficult and expensive to support, providing paybacks in terms of IT staff time and developer productivity.
This closed-loop process has reduced our response times from weeks to days. Prior to this virtualization initiative, developer requests could trigger a four to six-week negotiation process between R&D and IT involving extensive paperwork and multiple meetings. About 95 percent of the requests that we currently receive from R&D are satisfied within two working days. More than half are satisfied within a single workday.
Using the Four Keys to Virtualization Success
Using the four keys to virtualization success
The following four steps were critical to the success of our virtualization initiative:
Step No. 1: Standardizing offerings
Prior to this virtualization initiative, developer requests tended to be treated on a one-off basis. Individual requests could trigger a protracted negotiation process between R&D and IT-even though 80 percent or more of the required assets and configurations were common to multiple requests. We leveraged virtualization technology to deliver standard configurations with ruthless efficiency-and that efficiency has resulted in much shorter response times, happier customers and the elimination of the negotiation processes that occurred in the past.
Step No. 2: Allowing for headroom
Although the overall level of spending on server and storage resources will decrease through virtualization, it is critically important to purchase excess capacity in advance of demand. The effectiveness of our closed-loop provisioning process would be significantly undermined if we did not maintain the surplus capacity needed to respond to developer requests on a JIT basis. We routinely try to maintain 10 percent capacity in excess of current demand levels to ensure that we have sufficient capacity available at all times.
Step No. 3: Automating management processes
We used a BSM framework to initially design the overall set of processes needed to support this initiative. Then we employed system management tools in a surgical fashion to automate as many procedural steps as possible. With experience, we grew increasingly confident in our ability to eliminate operator intervention or supervision at specific stages of the overall process. We ended up automating steps in year two of this initiative that we would never have conceived of automating in year one.
Step No. 4: Enforcing policies
Our closed-loop inventory management process has evolved into a positive feedback loop. We are now able to provision virtual environments in a matter of hours or days and our R&D customers are comfortable checking resources in and out of our virtual pools because they are confident that their requests will be satisfied in a timely fashion. Also, management has provided the funding we need to maintain the necessary headroom capacity within our virtual asset pool.
It didn't always work this way. At the outset of this initiative, we needed management support to enforce a "virtual first" policy in responding to developer requests. We needed further support from management in enforcing asset reclamation policies when our monitoring tools indicated that deployed resources were being underutilized. Our positive feedback loop needed a couple of strategic shoves from management along the way to become self-sustaining.
Before you read another set of articles about cloud computing or attend another seminar on the topic, look into your own data center and determine how many underutilized assets you already own. Think about harvesting additional cycles or capacity from those resources through virtualization. The financial benefits of virtualization are well established by now, but if your experience is anything like ours, you will realize additional benefits in terms of IT staff productivity, customer satisfaction and time to market (TTM). Along the way, you will also be able to minimize server sprawl, avoid or defer data center expansion, and reduce your current levels of power consumption.
Proficiency in the use of virtualization technologies is similar to proficiency in athletics-practice and persistence will produce quantitative results that can significantly exceed your initial expectations.
Mark Settle is Chief Information Officer at BMC Software. Mark joined BMC Software
in June 2008. He has served as the chief information officer of four
Fortune 300 companies: Corporate Express, Arrow Electronics, Visa
International and Occidental Petroleum. Mark has worked in a variety of
industries including consumer products, high-tech distribution,
financial services, and oil and gas. During the early stages of his
career, Mark was the director of a systems integration business unit
within Hughes Aircraft Company. Mark is a former Air Force officer and
NASA program scientist. His formal training is in the geological
sciences. He received his Bachelor's and Master's degrees from MIT and
a PhD from Brown University. He can be reached at firstname.lastname@example.org.