Issue link: https://cp.revolio.com/i/288390
THE DATA CENTER JOURNAL | 15 www.datacenterjournal.com for scaling as the business requires. In addition, it may put higher-tier applications within the budget of companies that cannot afford, say, a full Tier IV data center but can afford a Tier IV–level block in a por- tion of the data center. • Better disaster recovery. In a multi-tier model, the boundaries of the more critical and less critical applications are clearly marked, enabling targeted disaster recovery. Personnel can focus on the higher- tier areas first, leaving lower-tier areas for later. Multiple tiers can also benefit colocation data centers that aim to serve a range of customers with different availability needs. Customers with mission-critical applications can opt for the higher avail- ability (and, naturally, higher costs) of a Tier IV–style service, while those who prefer lower costs can choose, say, a Tier I–style offering. MULTI-TIERED DATA CENTERS: MORE THAN ONE Multiple tiers can be implemented in a single data center or across multiple data centers. For instance, a company running several different facilities could dedicate one to mission-critical applications and the others to varying lower availability levels. is approach can run into difficulties if it involves large geographical distances between facilities and if latency affects cross-dependencies. Assuming applica- tions can indeed be divided among tiers, it is a design possibility for companies run- ning multiple data centers. CHALLENGES OF MULTI-TIERED DATA CENTERS Naturally, no design approach is all positive. Multi-tiered data center design brings a number of challenges that can eat away at the cost and efficiency savings out- lined above. In particular, the simple view of multiple availability tiers assumes that tasks can be neatly segregated: some tasks are clearly Tier I, some are clearly Tier II and so on. Although that may be the case for certain applications, others are highly interdependent. A particular application may by itself seem less critical, but if a mission-critical application depends on it to function, it effectively becomes mission- critical also. ese cross-dependencies increase the complexity of the design process. e following are several impor- tant questions to ask when considering a multi-tiered data center design: • Does the range of applications warrant multiple tiers? Clearly, if you have some applica- tions that require Tier IV–level availability and a majority that require Tier I–level availability, the multi-tier approach has the potential to greatly reduce costs. If your applications are mostly mission critical with a smattering of slightly less critical tasks, the cost benefits are less compelling. In such a case, the cost premium of a purely Tier IV data center could outweigh the benefits (and added complexity) of a multi-tier facility. • What are the actual cost benefits of multiple tiers? Don't go by estimates, which oen represent maximum savings in an ideal or nearly ideal case. Do the calculations with the design team to get a solid estimate of how multi- tier design can benefit (or not) your specific project. • How complex is division of the ap- plications among tiers? Spreading highly cross-dependent applications among different avail- ability tiers can be challenging, so research beforehand how difficult the project will become if you use multiple tiers. As with any data cen- ter project, failure to plan is likely to consume any cost savings you might be expecting from a particular design approach. • How will multiple tiers affect certifications? If certifications and other matters of public recognition are crucial to your business, it's important to consider beforehand how multiple availability tiers will affect your eligibility. For instance, if only a portion of your data center imple- ments the requirements for Tier IV, how will that affect your marketing capabilities? CONCLUSIONS Use of multiple availability tiers is an outworking of modularity than enables data center operators to reduce capital and operating expenses by only provisioning the infrastructure that particular groups of applications require. Supplying Tier IV electrical and mechanical equipment for tasks that require Tier I availability, for instance, is clearly overkill—and over- kill costs time, money and space in the data center. Naturally, each company will face different challenges and potential benefits in following a multi-tiered design approach. In some cases, a preponder- ance of applications requiring a certain availability may make the hassles of a multi-tiered design outweigh the relatively meager cost benefits. On the other hand, companies with relatively few mission- critical tasks can benefit greatly from this design approach, which only provisions enough supporting redundant infrastruc- ture to meet the needs of those tasks; less critical applications can then rely on less redundant or even non-redundant infra- structure, depending on their availability requirements. Of course, availability tiers need not conform to official Uptime Institute standards—these standards, however, provide a good basic nomenclature for evaluating availability requirements even if data center operators choose not to pursue formal certification. In a given design, different tiers can be defined arbitrarily as the company finds most beneficial to operations. Whatever the definitions—as long as they differ significantly in the amount of equipment required—designing to accommodate multiple data center tiers can improve the scalability, efficiency and flexibility of the facility. e key, however, as with any design approach, is to carefully evaluate the options beforehand in light of the goals, requirements and restrictions of the project. n