What sets it apart is that, unlike those other things, data center modernization never ends. It’s a continuous improvement cycle. Processes such as data center consolidation and virtualization should be looked at as steps along the way to modernization.
As you consolidate, you have to ask yourself what are you are going to do differently with your new infrastructure. When you do, you are talking about modernization, how to improve what you do with the data center, and how you are going to get more out of it. That is when the iterative process of modernization sets in.
They’re concurrent rather than sequential. As you reduce the number of physical servers to the optimal level you also have to decide on how to physically manage the IT assets that remain, which relates to the plant itself, the buildings the servers are housed in, the required energy consumption, and so on. All of that can be optimized and improved because you’ve also consolidated down to the appropriate number of data centers.
Besides the physical plant, management software requirements change dramatically in a modernized, consolidated data center. The density of computing resources and the fact that they’ll most likely be virtual rather than physical servers requires techniques and tools that are different than were used in the past. With the modernized data center you will find a very strong return by using the appropriate modern tools.
There are general metrics that will apply in all cases, but the actual numeric targets will differ depending on the agency or the mission specific organization we are working with. One simple metric could be the ratio of administrators to the number of virtual machines, for example, and another the number of storage area networks that are managed by each administrator.
If you look at those, for each case you can decide whether you need more or fewer human beings to achieve the increases in productivity that meet the organization’s objectives. And that will speak to how you deal with modernization, because the tools you give an administrator to manage a thousand virtual servers are very different from those they would use to manage 50 physical servers.
Another approach would be to map the goals for modernization to how quickly the data center could address increased capacity requirements. That would show how resilient the organization would be. It can be measured in how long it takes to provision new server instances, and can be used to develop appropriate service level metrics.
What the tight budgets and actual budget cuts mean is that the upfront money needed to pay for that initial transformation to data centers and IT assets is very hard for agencies to come up with right now. Even though you know there’s a long-term benefit with modernization, and you can expect a true reduction in costs, it’s the cash flow they need to do the transformation that’s stopping people.
One thing you can do in this situation is to change the financial model. For example, many state and local governments are looking to an external agent such as an integrator or contractor to put up the seed funds for these projects. The external agent then can recoup some of the shared savings from the modernization across the out years. Or agencies can enter into formal public/private partnerships to do the same.
Currently, the most significant technology is virtualization. It is a foundational aspect of so much in data center modernization because it leads to very flexible networks. We are starting to gain the ability to move compute resources, whatever they are, to better respond to demand. Without virtualization the speed at which you can provision and de-provision resources is greatly constrained.
In order to accommodate any future technologies that would likewise help, it’s best if IT leaders adopt more of a portfolio approach. As part of that they can have some technology long shots along with the low-risk meat-and-potatoes stuff, and with that they’ve got to take on an early adopter mindset and know that at least some of the technology that’s brought onboard is going to fail.
In the public sector that’s a very hard thing to do, I know, because all the successes are expected and all of the failures are publicized. That’s what makes adopting new and high-risk technologies such a potentially painful activity. However, if you ran a business this way, you’d still be trying to sell coal from the back of a horse cart. The risk-free way just doesn’t work anymore.
Most of those decisions are being driven by policy rather than technology. If you look at some of the reasons why people prefer colocation to the cloud or to shared infrastructure, they’re really predicated on what the machines are doing, what the security requirements are etc. It’s the data itself, and what it’s being used for, that’s driving the choice.
Certain customers will not tolerate a shared infrastructure, primarily because of security concerns. There are also other benefits to being in a modern building, such as fire suppression and redundancy of power supplies. You’ll get that from colocating in a modern facility.
More than anything, it has to be a collaborative effort about what’s technically possible at what cost, how that maps to the mission objectives, and how it all agrees with the capabilities and the financial option that the government agency puts on the table. It’s very difficult to say there’s one right way of doing modernization. There have been a lot of people that have tried this, and there’s no lack of lessons learned.
The one element people need to look for is flexibility. Match the potential of the modernization with the mission and then start the discussion about where to go with it. And remember, there are right ways and wrong ways to do it, but you have to know what you are trying to achieve at the outset.
SOA is a technique for describing a more flexible and resilient infrastructure, and it’s a component of data center modernization because it provides a framework or a language we can use to describe IT processes. There are other frameworks that people are using in IT that can similarly be employed for modernization, as long as the one you use fits what you are trying to describe.
The point is that, whatever you choose, you need to stick with it long enough that it can help change some of the decision making. More often than not, people do SOA or something else for a bit and they think it’s fun and they go to offsite retreats and think deep thoughts, but it ends up not doing anything for them. SOA and modernization needs to be embedded in the culture, as does the notion of continuous improvement.
This is when we start to talk about governance and the kinds of intangibles you can’t find on the data center floor.
Governance is about the rules that are needed to manage modernization and who gets to change them. That’s important because, if you build something that’s flexible by nature and by design, it will change over time. So governance has to be put into the planning process from the get go.
Any technology you use in the modernization process will eventually become obsolete. But if the flexibility of the decision making process and the dynamic between IT and the various lines of business in the agency are set up properly, then the chance of success of the modernization is greatly enhanced.
Because of financial constraints some agencies are trying to defer and delay server and software upgrades. So, to accommodate modernization needs, I’d say that, if an agency has a substantial IT infrastructure, it would be better for it to use a recurring hardware cost model such as leasing, rather than to actually own the assets. For software, move to a pay-for-service mode. Try and reorganize assets away from ownership and into usage so that you have the greatest possible level of flexibility.
Many of the agencies we do business with are very open to this. We have many customers who are approaching us to restructure enterprise agreements away from perpetual software licenses to ones based on user counts or managed objects or some other unit of measure. Anything can be put on a shorter refresh cycle if you change the financial model that you use to pay for it.
Designing your infrastructure, especially your applications, to be much more flexible when it comes to end user devices, is one of the foundational changes that has to be made. People are expecting to have the same application available to them on tablets, smartphones, laptops and other devices. Data center modernization can allow you to address changes in workload by, for example, moving end user devices off the physical premises.
However, the idea of portability of applications doesn’t come from modernization by default. Having an end user device outside of your direct control is something you have to design for. Even though you won’t control everything anymore, you still have to provide a robust and meaningful computing experience for the end user. You have to design things to take into account all the possible devices those end users will employ and, more importantly, you have to change your mindset to accept that that will be the way going forward. If you don’t, you’re sunk.
The real start is a mindset that you’re going to design something that will adapt, which goes back to the idea that data center modernization is something you never finish. So you have to design into the system processes to handle ongoing, iterative change.
Again, the first thing needed is the governance structure. If the rules for change and how change is going to be accomplished are known upfront at the beginning of the modernization, the acceptance of change over time will happen more freely. The best system is one that is consciously designed for constant turnover.
When I think of the costs of computing I think of them as broadly defined, such as the physical plan, hardware, software and people. It’s only relatively recently, since energy prices got really high, that energy costs have become important. Before that, no-one was all that interested, they just turned up the air conditioning.
Obviously, if you’ve got a more modern building it’s going to be more energy efficient. But I’ve seen data centers put into old cannery buildings, old school rooms and recycled shopping malls, which are not nearly as efficient, and which puts the capital expense at a very high level. Because of that, state and local governments in particular have started leaning very hard on integrators to fund new data center facilities that are more energy efficient.
Data center modernization now calls for those kinds of buildings.
You can put a software layer on top of your existing infrastructure to achieve some gains; to optimize what you have without having to throw everything out. Software to improve server utilization, for example, so that you don’t have to buy more physical servers. That has an impact on heat output, energy costs, and the size of the facility you need.
There are three different pillars that support modernization. One is technology, so you need to include someone who has a very good technology management background. Then you need financial expertise because you must have a clear idea from the beginning from where you are going to get your funding. The third, if the IT organization is truly independent of the agency’s lines of business or mission, is that you need people at the table who can speak for the business side of the agency. We’ve found over the years that, if you have only two of these three pillars in place, the modernization will fail.
Is this understanding widespread in agencies? When IT was seen by the business side as more of a service provider, it used to be “OK, Mr. Geek, sit in the corner over there and I’ll tell you when to talk.” That’s changed in recent years. Now, IT is embedded in a cabinet person or a political appointee, in the position of chief information officer or chief technologist. And that really stresses the importance of technology today, as an equal partner in the organization’s fortunes.
If an agency today still relegates IT to a lesser position than the lines of business, then it is not set up to succeed for data center modernization. Modernization really needs to be a true three-way partnership between finance, IT, and the lines of business.
8609 Westwood Center Drive, Suite 500, Vienna, VA 22182-2215 703-876-5100 © 1996-2016 1105 Media, Inc. All Rights Reserved.
8251 Greensboro Drive, Suite 510, McLean, VA 22102 | 703-876-5100
© 1996-2017 1105 Media, Inc. All Rights Reserved.