Preparing for the data center of the future

Shutterstock image: black data center with blue, glowing lights.

We seem to continually look to the data center for new ways to cut costs, and that is not likely to change anytime soon, given the ongoing attention and focus on consolidation. Indeed, the next five years will bring some dramatic changes in that area.

The data center of 2020 will look vastly different from today's data center in a variety of ways. As application silos are broken down and resource tiers consolidated, the result will be data centers that consist of three hardware tiers -- for processing, memory and storage. Applications will dynamically allocate resources from each of the tiers, providing the required elasticity to respond to changing demands.

With the advent of cheaper memory, more federal agencies will adopt in-memory computing technology to reduce application response times. That approach has the added benefit of transferring the load from transactional databases, which can further reduce licensing and operating costs.

In other words, the data center of 2020 will be able to do much more with much less by using resources more efficiently and having better insight into the use of, and control over, those resources. Furthermore, a well-managed data center will reliably support ongoing services and sustainably manage and ensure IT's contribution to the agency's mission.

So what should we be doing today to move toward the data center of the future?

1. Plan for incremental consolidation and show return on investment as you go.

Big-bang consolidation efforts tend to fail because of budget overruns and a diminishing sense of priority over time. At first the promise of cost savings is enough to secure substantial funding for data center consolidation efforts. However, as time goes on and deadlines are missed or priorities change, funding dries up.

For a data center consolidation to be successful, you must show incremental success with positive ROI. Start by tackling quick wins and take on bigger projects as you gain experience.

2. Understand your data center's interdependencies and use that knowledge to drive consolidation.

Most data centers are a tangled mess of interdependences, with applications sharing databases, directories, operating systems (or virtual machines) and hardware. It is essential to understand and document all the dependencies across application layers and tiers so that changes don't cause business interruptions. Without that understanding, migrating one platform could affect other applications and cause unplanned outages.

3. Catalog all aspects of your data center, from business process drivers to hardware.

Cataloging is an imperative step in any consolidation effort. Without a proper catalog of which applications, platforms and resources each business requirement is tied to, it is exceedingly difficult to plan a successful data center consolidation.

By investing in a portfolio management tool, agencies can document all business processes, the corresponding requirements and the platform stack components necessary to execute them.

4. Get the right team in place.

At the heart of any effective and efficient data center is a team of skilled and experienced professionals. That team generally consists of government employees and contractors with a variety of skills, all of whom are well versed in operating data centers and migrating applications.

First and foremost, you need planners who can work with business owners and stakeholders, an operations team that understands the priorities of the agency and any consolidation efforts, business analysts who determine which business functions are essential, application architects, infrastructure architects, and technical writers to update all the corresponding documentation.

5. Build a detailed plan.

Planning is the key to successful data center consolidation and operations. Whether you are simply consolidating or also migrating to the cloud, you must thoroughly understand the platforms involved and create a detailed plan.

That level of planning requires modeling your existing and "to-be" architectures down to the smallest details. It is even more important to tie together all the pieces -- from the business processes down to the serial numbers of the servers that execute those processes. That approach allows you to visualize the dependencies between all your platform stacks and run "what-if" scenarios. Best of all, it allows you to associate costs with all the dependent platform entities.

6. Establish a governance model.

As you consolidate, document and clean up your data centers, proper governance can ensure that they continue to meet the needs of stakeholders into 2020 and beyond. Establishing a governance framework can ensure that all changes are coordinated and communicated, requests for duplicate services are eliminated, licensing requirements are met, and solid architecture principles are followed.

It is also important to continue evolving the enterprise architecture guidelines in a data center that is purpose-built with change in mind. Proper EA tools will help ensure future auditability and the recordkeeping necessary to understand who changed what when.

7. Don't forget the security.

When consolidating infrastructure, it is important to understand the security requirements for each component and ensure that elements with high security requirements are not combined with those that have low security requirements.

You should also accommodate storage encryption requirements, facility location and IT acquisition review considerations, and auditability. Solid EA tools can help you plan the life cycle of your components so that all elements are eligible for support from original equipment manufacturers. Weak links in a consolidated environment can have a ripple effect on cybersecurity.

Data centers and the solutions hosted within them are never static. Even the leadership and priorities that we must adhere to and support can be a moving target. But by preparing today, we can ensure a more efficient tomorrow for the government's data centers.

About the Author

Chris Borneman is vice president of Software AG Government Solutions.

Rising Stars

Meet 21 early-career leaders who are doing great things in federal IT.


Reader comments

Tue, Jun 2, 2015 Chris Borneman

Thanks for your questions. Regarding the quote and the tiers, we are seeing a strong benefit towards memory as a form of storage, used for high performance caching. The cost of commodity RAM continues to drop, while the need for faster access to data with in-flight transactions continues to grow. This is very difficult to achieve with traditional database approaches which rely on storage. I also would expect SANs, leveraged with flash and hybrid devices, will continue to expand vs. everything encapsulated into one box. Virtualization benefits are enormous, and whether you focus on private or public cloud utilization, the dynamic allocation of resources from the 3 tiers discussed will be the standard for new server provisioning. How data centers are built I think still will continue to shake out. I don't know that containerized data centers will make the most sense for all cases, but repeatability through modularity seems to have its potential. The approaches for maximum energy efficiency vary, and very much come into play as part of a set of systems at design time vs. retroactive refitting once built. I think the driver will be the approaches for cooling and which approach of flooring will complement cooling trends will dictate what that will mean for raised floors.

Sat, Apr 18, 2015 Ryan Hulland

Chris, interesting article. I was especially interested in this quote: "...the result will be data centers that consist of three hardware tiers -- for processing, memory and storage." I'm not exactly sure how to interpret that, and would love if you could reply and elaborate a little bit. My first take on that is there will be more segregation between the 3 main hardware types....? Is that what you mean by that? Are you saying SANs and the like could make a major comeback, and the standard servers that are so prevalent today will fall by the wayside (servers with a HDD, RAM, CPU, I/O, etc. in one box)? I'd also love to hear your take on the basic infrastructure of the data center itself. Will modular data centers continue to grow in popularity? Will access flooring still be used? (This one hits close to home as my company, Netfloor USA, manufactures raised access flooring. So far, we've seen an increase in the use of this flooring, even in some of the mega data centers, but I know there are always the naysayers who say there won't be any use for raised floor in the future. Thanks! Hope you see my comment and can reply.

Fri, Jan 9, 2015 Chris Borneman

To the poster of the first comment, I guess I need to make sure my point there is made a little clearer. Nothing about getting the right team in place is meant to ignore the people who are already working in the data center. I have operated data centers for most of my career and absolutely had confidence in my data center team and leveraged them as much as possible. The point here is that when you look at consolidation, the best approach I have found is to have an extended team, involving the application owners and business analysts who also understand what those applications do and which functional interdependencies exist. You will also want to scale up the planning team and make sure communications are early and active with all stakeholders that will be impacted or benefit from the consolidation exercise. This extended team of course incorporates the existing data center team, not replacing it.

Fri, Jan 2, 2015

4a) Remember that outsiders always know better than any of those worthless people who already work in your data center.

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group