The Design and Organization of Data Centers/Introduction
Intended audience[edit | edit source]
| This section is a stub.
You can help Wikibooks by expanding it.
Introduction to the topic[edit | edit source]
The design and organization of your company's data center has a direct impact on its day-to-day operations. The following are just some of the considerations involved:
- Capacity. Your data center's capacity needs to be optimized in order to ensure that it can handle its workload. In this regard, there are three common approaches to capacity planning:
- Start with the most economical size then expand it, careful not to increase capacity beyond actual need.
- Start with the size that best will be able to handle the projected initial capacity, then expand it as need increases.
- Start the data center with a size that will handle the highest possible load that could conceivably be initially placed on it, and expand vigorously to keep ahead of load.
- Expandability. The data center should be scaleable so that if you choose to increase capacity, you will be able to do so without having to redesign the entire data center.
- Uptime. You will need to determine what level availability you will need. A data center that is designed for 99.99999% availability(3 seconds of outage per year), for example, will require extremely careful planning, as well as a commitment of resources that would not be necessary if you only are aiming for 99.99% availability(52 minutes of outage per year), and yet even a 99.99999% available system may not be adequate for a life safety application, where a fraction of a second of outage could cause a severe risk of harm. It is important to keep the availability requirement in mind during the planning and design of your data center.
- Length of outages. Closely related to uptime, you will need to determine the maximum acceptable length of outages. For an E911 service, for example, 1 second of outage would be 1 second too many; for a business-critical service, you typically will perform a cost-benefit analysis. Length of outages is typically quantified as mean time to repair.
- Investment. How is the data center financed? How much money is available? How is it best spent?
- Location. Will the data center be located at your corporate headquarters, or off-site? Will you rent or own the location? Will the data center be co-located, for example, in a telephone-company central office or a carrier hotel? You will need to think about considerations as diverse as the cost of communications network connections, the cost of providing benefits to any employees that may work in the data center, and the legal environment.
The need to deliberately design and organize[edit | edit source]
To maximize uptime, minimize length of outages, and optimize capacity subject to resource constraints, one should deliberately design and organize the data center. The following considerations merely highlight the importance of careful planning:
- Large investment. When you consider the costs of network connections, power and utilities, and rented or purchased space--to name only a few things--you can see the cost of starting a data center is more than the cost of computers and network equipment alone. After you start the data center, there are still the costs of operating it: the annual operating costs  of a data center can easily reach into the tens of millions of dollars. It is important to optimize the allocation of resources to get the maximum "bang for the buck."
- Hard to change. Once a data center is up and running, it is very hard to change its design without bringing the entire data center down.
- Hard to fix. Generally, design problems are expensive to fix after the design has already been implemented. Thus, if your customers experience a problem that is inherent to the design of your data center, the cost of repair may be prohibitive.