In late June, as President Barack Obama was directing federal agencies to consolidate their data centers, the commonwealth
of Massachusetts broke ground on a $110 million facility in Springfield, Mass., that will help it do just that: consolidate 183 data centers into two mirrored facilities housing a total of 4,000 servers at full capacity.
In early 2009, Massachusetts IT and business leaders, already deep into the data center planning process, determined that the commonwealth's massive patchwork of technology -- with 100 phone systems, 24 email systems and 15 data networks -- was unsustainable, and named IT consolidation as the No. 1 priority in its strategic plan. Supporting the consolidation is the new 145,000-square-foot Springfield data center, which will be paired for backup and disaster recovery with an existing Massachusetts Information Technology Center data center in Chelsea, Mass. Stuart Lecky, Springfield data center program manager; and Michael Johnson, enterprise architect, shared their approach to data center planning, their vision and their strategy for the dual data-center build-out.
Why did the commonwealth decide to build and maintain its own data centers and not work with a collocation or cloud provider?
Johnson: The main reason is for security. With your own data center, you can control access carefully and completely. Some agencies and [our CIO] secretariats also have regulatory requirements for physical separation of data, even from other state agencies. Secondly, one of the primary aspects of the two-data-center model is disaster recovery. In the case of a disaster, the remaining data center needs to have the ability to restore and present all data from the site that is no longer available. Our infrastructure is very complicated, with many systems that have to be replicated. It would be impractical to work with another party to develop recovery scenarios for the systems and meet the recovery time objectives that need to be supported. We need to control the infrastructure in order to create and maintain the redundant systems required.
Lecky: Also, we are under way with Executive Order 510 to consolidate IT resources. The key is the consolidation. We received capital funds through an IT bond issued by Massachusetts, so the IT organizations are not paying back lease costs or capital debt payments compared to private industry. The selection of Springfield satisfied the geographic diversity requirement of the funding legislation, which called for a site in western Massachusetts, with a separate network, power grid and IT infrastructure.
Michael, in your blog, you mention that the Springfield data center uses Energy Star servers. Could you describe other ways you plan to increase energy efficiency?
Johnson: In addition to the facility innovations, the technical infrastructure will help support the energy-efficiency goals. By using new technologies such as [10-Gigabit Ethernet], we can significantly reduce the amount of cabling, network ports and network devices within the data center. Innovations in storage, such as solid-state drives and auto-tiering, will also contribute to energy efficiency.
Although we plan to implement best-practice new technologies at Springfield, we have to assure that those systems will be interoperable with systems at our existing data center in Chelsea.
Michael Johnson, enterprise architect, Commonwealth of Massachusetts
Lecky: The facility is designed with many energy-efficient technologies. One of these technologies, air side economizers, utilizes cooler outside air to cool the indoor spaces. The commonwealth's division of capital asset management and information technology division have also been working with Western Massachusetts Electric Co. to maximize potential rebates available through the utility, to lower front-end costs. We expect to cut energy bills by 50%, saving $2 million to $3 million annually. The Springfield site will be one of the commonwealth's top consumers of electrical power, at 6 megawatts at full capacity -- that's 15% to 20% of the current statewide contract for office buildings. We are working with the green construction industry and have posted our plans on our wiki.
How is virtualizationused in your data center planning strategy?
Johnson: Virtualization will be the main contributor to energy efficiency. We plan to drastically reduce the number of physical servers through virtualization. By using features of the virtualization software, such as resource scheduling and power management, servers can be powered off and on automatically as required, while performance can be guaranteed. These features will help achieve high rates of utilization and efficiency.
How will automation play a role?
Johnson: We're working on an enterprise automation tools strategy that brings together all the areas of the enterprise that need business services management. This includes event management, asset management, monitoring and more. The Springfield data center will be a lights-out facility, with minimal staffing, so the need for automation is critical. Our long-term goal is to provide services and resources through a virtual private cloud. This will only be possible through careful integration of our tools strategy.
What are the challenges of the consolidation?
Johnson: We had to start thinking and planning more from an enterprise perspective. Although we plan to implement best-practice new technologies at Springfield, we have to assure that those systems will be interoperable with systems at our existing data center in Chelsea. We are currently working on our enterprise virtualization project to assure we are ready for things such as replicating virtual machines and other data between data centers. This requires a great deal of planning and coordination. Other enterprise initiatives under way are storage and backup and recovery.
Do you have advice for other organizations consolidating that many data centers?
Lecky: Get an excellent inventory, and develop a repeatable process for your migrations to a virtual environment, from physical machine to physical, physical to virtual, and virtual to virtual.
What aspect of the consolidation are you working on now?
Lecky: [We're] looking at lots of standards for server, storage and networking. We haven't determined all of those, but Fibre Channel over Ethernet is a good one.
Johnson: We're working on developing a long-term enterprise storage strategy, exploring and procuring backup and recovery technologies [which comprise a lot of products]. And we're working on the virtualization strategy.
Let us know what you think about the story; email Laura Smith, Features Writer.