Finding the Right Data Center Model for Your Business

James Young is Technical Director, Asia Pacific at CommScope.

There’s no question: Data center needs change as business applications change direction, grow or die off. Managers can align their data center capacity, choosing to stay with their own onsite capacity or migrating all of their applications to a public, shared cloud. Some also may choose to rent space in a multi-tenant data center (MTDC), moving away from their own facilities. There’s also a happy medium in which businesses mix all of these alternatives into their own unique blended solution. So how do data center managers decide when and how to evolve their data centers? Naturally, there are pros and cons to each data center model.

On-Site Data Centers

In this scenario, managers control their data center at their own location. An on-site data center improves efficiencies for some business needs. Of course, it also carries maintenance requirements. With no crystal ball as to what will happen in the future, it is more difficult to scale this investment up or down: guess wrong, and the cost of this alternative can be much higher than other choices.

All of the activity around cloud computing makes it seem like on-site data centers are dinosaurs, but there are a couple of different reasons why companies choose to own their own data centers. For example, some companies have static requirements and perform a significant amount of processing on an ongoing basis, and they have invested in the data center capacity to do that. Over time, they might make changes to the data center, but it ends up being more expensive to go into a leased facility or into the cloud unless there’s a good reason to do it, such as changes in operations or technology.

Insurance companies and banks, for example, have been using the same applications for a long time – they’re part of the organization’s core capabilities and are considered to be a strategic advantage. There are also companies that work with large data sets, like oil and gas firms. Moving those large data sets around to different locations – like into the cloud – is very expensive and time-consuming.

Also, many enterprise workloads aren’t built or designed for cloud or can’t be virtualized, such as applications written in COBOL for mainframes. Changing those applications and replicating that software in a different environment is a very large and expensive undertaking. At some point, of course, legacy applications such as this will end up being rewritten to run in virtualized or cloud-native environments when the needs justify the expense and risk of doing so. For example, financial services firms that run online transaction processing or actuarial applications will probably take quite a few years to rewrite these applications based on the higher risk of change and the complexity their own customized solutions contain.

Cloud Computing

From the enterprise perspective, the trend definitely appears to be toward increasing use of cloud over time. When an organization moves its workload to a cloud environment, it’s different than simply converting to virtual machines (VMs). Enterprise data centers support many different applications, and this can translate to hundreds or thousands of VMs. It’s difficult to manage this new virtualized environment while ensuring high security and availability. Adding automation and orchestration tools greatly enhances operational efficiency and can turn multiple VMs into a private cloud. Now resources are agile and available for whatever workloads the business may need to deploy.

Private and public cloud environments differ in several ways. When a company chooses a private cloud environment, the company has the benefit of absolute control. Internal operating costs may be much less than the monthly charges from using a public cloud depending on the way the data center services are used.  Evolving to a private cloud is a lot easier from a security and management perspective than moving to a public environment; it is common practice for the enterprise for the most part.

The public cloud is a rental environment that offers much of the same facilities we see in private cloud. Here, the next wave of applications is typically rewritten as cloud-native applications to run on specific types of public platforms. Public cloud moves companies completely out of the infrastructure business, and it might offer better security than small or mid-sized enterprises can manage on their own.

But regulations can be an issue: the regulatory environment hasn’t necessarily caught up with what’s possible with public cloud. For some companies, there are several regulatory hurdles to overcome before moving to a public cloud. That’s why many banks, for example, often host private clouds in their own facilities.

Another downside to public cloud is that there’s no standardization across public cloud platforms. Cloud vendors have unique ways of doing things. It can be difficult to change to another vendor down the road. Putting software into a different cloud environment can be troublesome and expensive. If you have very large datasets in the cloud, it can be hard and expensive to move them into another facility.

Cloud also forces managers to give up a certain amount of control in terms of service level agreements. If a public cloud fails, managers have to be aware of backup and recovery plans. Once he or she has made a commitment to the public cloud, it may be impossible to move back.  Will you still have resources to manage a private environment? If you recovered your data, where would you back-up applications run?

Still, if your company isn’t using any differentiated platforms, chances are the cloud-based applications are good enough, and they may be easier for customers to access. But large enterprises that may have invested millions in developing proprietary applications, which they use to differentiate themselves, are not going to cloud in droves.

MTDCs

MTDCs offer the ability to pay for infrastructure as a utility rather than running it in your on-premise environment. Even complex enterprise resource planning (ERP) environments with millions of dollars tied up in customization might not justify building and owning a private data center. In this case, it may make sense for them to move it into a hosted facility and buy space, power, cooling and connectivity from the MTDC. And, since it is commonplace for major public cloud, service and content providers to also have a presence in MTDCs, enterprises can connect directly to them.  This can significantly reduce latency and user experience, and simplify their planning by having their public and private clouds, and carrier connectivity, under a single roof.

Choosing Models

Finally, a lot of companies are using more than one data center model. A company might use public cloud to enable self-service access to compute resources, for example, while it runs proprietary, business-critical applications in an on-site data center or an MTDC.

In the end, different companies have different needs, and they use different data center models to meet them. There are several alternatives available today, and you can customize the investment strategy and the migration path to implement a combination of hardware/software approaches. Cloud is becoming more popular, MTDCs are becoming much more capable, and regulators are trusting off-premises solutions more. The trick is to use the best model for the task at hand.

Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating. View previously published Industry Perspectives in our Knowledge Library.