Today the industry is suffering from a dot.com “hangover” from the old days when data centers were designed to be big, beautiful, highly redundant and complex. For instance, many end users may think they need a Tier III data center at 300 watts per square foot. For most, this is simply too much redundancy and capacity (avg. user is Tier I-II and about 150 W/ft2).
Likewise, traditional critical facilities typically house all the support equipment (UPS and Distribution) inside the space. If you are a cloud or hosting provider, your primary challenge with this brick and mortar approach is that you're eating up rentable IT space with support equipment. This type of design also often times:
As power prices continue to soar, energy consumption regulations tighten and technologies such as cloud computing become more widely adopted, organizations will be forced to find more efficient and competitive ways to supply computing power. However, we see the same stale building methods continuing to be applied over and over again, hindering organizations' growth and leaving problems unsolved.
Challenging the Establishment
So, how do companies move their data centers into the future and reap the benefits today? The first step is to change the mindset in which you think about data centers.
Companies must start thinking about the total cost over the lifespan of their facilities, and plan accordingly. This conversation needs to start before the first design is drawn, and before any concrete is poured. One way to bring these costs down is to stop designing for tier levels, and start designing for "concurrent maintainability." By applying this concept, it is possible to maintain the same uptime levels with a lower capital expenditure and lower operating costs without traditional tier designs.
To further maximize your organization's return on investment and lower total cost of ownership, we recommend following the Four Principles of data center construction:
1. Preserve capital - First, avoid spending as much capital as possible to minimize financial investment. Build incrementally at the lowest capacity needed, stair stepping the use of capital in a just-in-time manner. Build as simply as possible, focusing on concurrent maintainability to meet uptime requirements rather than robust redundancies.
2. Minimize operating costs - Building less upfront and lowering the tier level will automatically lower operating costs. But you must also take into account the design itself. The simpler the design, the more efficient, green and maintainable it will be.
3. Be flexible - Stay as flexible as possible and don't build yourself into corner. Keep as many elements open as you can, such as physical space and redundancy, by building with a Lego-like approach. This will not only bring costs down now, but also allow you to grow cost-effectively in the future.
4. Measure - the only way to tell the effectiveness of a strategy is to measure the results. This is actually a two part process. First, measure before the build by modeling costs on all sides of the facility throughout its lifecycle. Second, measure after the build is complete and throughout operation to know its true effectiveness. This will help create a competitive advantage.
By following these four principles, an organization can develop a data center strategy that addresses budget, time to market, redundancy, maintainability goals, preservation of capital and lowering operating costs.
One solution that is gaining in popularity across the industry is a containerized or modular approach. Containers and modular designs may be a highly viable option for some because they significantly lower the upfront capital expenditure to get into a space. Modularity lowers all levels of costs (build, operate, maintain, etc.) and containers or modules are assets that can be moved with you. Containers also provide extreme flexibility with location and concurrent maintainability levels, which can address the needs of everyone from cloud providers, hosting companies, corporate clients, hospitals, universities, and others. The possibilities are virtually endless.
On the illustrations below are two examples of how modularity can be used as a business strategy and its benefits.
The first model illustrates a partially modular application with traditional raised floor space where all facility support systems have been moved outdoors in weatherproof modules or containers. Assume that this is colocation space and the company isn't certain whether to build a Tier I, II or III space, but knows they need something to show perspective tenants.
For the technical guys, this illustration represents an 'n+1 configuration' with a UPS module, generator module and distribution module. In this scenario, this company not only frees up valuable revenue-generating floor space, but they also lower the capital cost by being able to simply add modules to increase capacity or redundancy. Instead of building the entire infrastructure from the ground up, burning capital and throwing away cash needed to maintain operations, this colocation/hosting company can build quickly and efficiently as capital (and the client) requires.
The same concept applies to the IT environment, only the benefits are magnified significantly. This is especially true if you design the IT modules or containers using the Four Principles.
While there are many options available, it's important for you to determine what's right for your organization. Modular is certainly not always the best approach for every need. Think about the total cost to own and maintain your facilities. Consider what you really need and set realistic expansion requirements. Ignore tier levels, cost per kW and the lemming in front of you, and craft what fits you first. This is the best way to ensure a successful and efficient data center program for any organization.
If you're interested in learning more about modular or containerized solutions for data centers and how they work, please see our free whitepaper Evolution of the Data Center: How to Use Modular Designs to Gain a Business Advantage at www.leetechwhitepapers.com or contact us.
Article published 10th January 2011