Buying data center solutions is a lot like buying a car: there are so many options, it becomes difficult to choose the right features to fit your needs and your budget, and renting a car or taking a cab might make more sense. You need to determine what combination of hardware and software you need for the job and what kind of infrastructure you will need to support them. You also need to consider whether it’s more cost-effective to host the entire data center yourself, find the right colocation partner, or to adopt some kind of hybrid solution that may include hosted and cloud resources. You want to be able to make well-reasoned comparisons of similar solutions in order to make sure you assemble a data center that can meet your immediate needs and that can adapt as your needs change.
Installing and maintaining any type of data center is an expensive proposition, but how do you gauge costs? Whether you plan to build or lease a data center, you need a common metric to gauge expenses. Most data center architects assess capital expenditures (CAPEX) for the outlay for hardware and infrastructure and operating expenditures (OPEX) for power, facility staffing, maintenance, and other costs. The challenge is that the level of complexity makes it almost impossible to project CAPEX and OPEX costs accurately. Servers are likely to take up most of the data center operations costs, followed closely by energy costs. However, if you use virtualization and other strategies, you can substantially reduce server costs. Energy costs can be reduced with new cooling strategies and smart power monitoring. And the CAPEX/OPEX formulas don’t work when you integrate cloud or colocation services.
So how do you get an apples-to-apples comparison to assess your best data center options?
Hosted Versus Cloud Costs
When assessing in-house versus cloud computing requirements, start by balancing the benefits of outsourcing computing resources and the savings in CAPEX versus the OPEX of using hosted data services. If you are looking to get the data center operational fast, assess requirements, and build as you go, then starting with a cloud solution can make sense. With the cloud, you can reduce up-front costs and use a pay-as-you-go model, plus you get extensible resources that can expand with demand.
There are obvious advantages to using cloud computing services:
- Service is available on demand, and there is scalability with almost infinite capacity
- You can access resources from any location
- Infrastructure abstraction to applications isn’t locked to specific devices
- You gain consistent service-level characteristics regardless of computing load and
- Cloud services have a usage-based billing model with no initial hardware investment
Of course, there may be business or legal considerations that may make a hosted data center impractical. For example, is there concern about protecting intellectual property and security? Will sensitive data be hosted in a foreign country? Do you need to meet protection and legal requirements for regulatory compliance? What about disaster recovery? Any of these factors could be a deal-breaker for hosted services and give you sufficient reason to build your own data center.
When comparing costs, as part of the apples-to-apples comparison, you will need to compare the cost of in-house data center solutions versus the monthly fees for hosted data center services. That means determining the cost of hardware (HW), software (SW), staffing (SS), support (SU), facilities costs (FA) such as electricity and building management, and any other costs. To determine the costs of maintaining your own data center, you apply the formula:
Once you have an estimate of the operating cost of maintaining your own data center, you can better determine if outsourcing is cost-effective. There are going to be complexities in managing your own data center that may be difficult to factor in, such as changes in demand that could affect CAPEX and OPEX costs, but it’s a good place to start.
Can You Save with Colocation?
If you want to consider using a colocation facility instead of an in-house or hosted data center, then you have a different set of considerations. Colocation offers a different set of advantages, giving you a hardened facility that provides better uptime, better security, and better protection from natural disasters as well as a choice of telecommunications carriers. You will still have the capital expenditure of providing the necessary hardware, but the colocation provider typically takes care of basic elements such as floor space, rack space, power, managed services, and telecommunications.
Using the same approach outlined above, you need to determine your in-house data center costs for those same services. The big-ticket items are going to be power, managed services, and telecommunications.
Assessing power is a matter of assessing your hardware needs against the colocation provider’s billing model. Is power billed by actual usage, or is there another formula? If downtime is a concern, be sure that the colocation site has redundant power, generators, and reliable uninterruptible power supply (UPS) systems and be sure to include those hardware and operating costs in your cost comparisons.
Managed services are another consideration. Is it more cost-effective to use your own IT staff or outsource managed services? Do you need remote hands or smart hands for support? Some colocation providers bill support whether you use it or not, and others pre-bill hours. Make sure you understand support costs for comparison.
Telecommunications is one area where you typically will have more options with a colocation service. You may need point-to-point links, more Internet bandwidth, or a combination of circuit types. Assess the best telecommunications bundle to suit your needs and compare costs from the colocation provider versus installing similar services in-house.
When assessing your data center needs, using an apples-to-apples comparison can be challenging unless you understand how to deconstruct the basic cost factors. If you can break down costs to their most basic levels, it should be relatively easy to see where your most cost-effective data center options lie.