As long as data centers are operating well, they tend to go unnoticed in an organization. At peak performance, data centers are maintaining optimal balances between server and storage technologies, space, power usage, heat produced and the cooling necessary to prevent overheating. Automation techniques even make it possible for data centers to perform a range of procedures on their own, such as simplifying maintenance and freeing the IT department to focus more on business-driven initiatives.
However, with many organizations changing or adapting their strategy to the market, data center requirements are also changing. Data centers in need of attention may have different types of signs. For example, physical signs include numerous displaced cables, different types of servers in and out of racks cooled by a bunch of fans and portable air conditioners. These signs indicate a lack of an evolving strategy.
Pressures are mounting for IT departments. Mergers and acquisitions are jamming more technology into tighter spaces. Increasing computational demands create crowded equipment racks. Staggering power and cooling requirements cripple budgets. In general, a high percentage of data centers are reaching the end of their lifecycles and becoming increasingly expensive to maintain.
In addition to the risks organizations face in terms of service interruptions and outright failure, they are also missing opportunities to leverage technical advances that can be critical to maintaining a competitive edge. Much documentation is available about savings in dollars and energy as a result of a strategy for server and storage virtualization and consolidation. But those savings cannot be fully realized unless corresponding changes are also implemented in the data center.
What about the so-called “greening” of data centers? Cooling accounts for as much as 40 percent of the cost of powering a typical data center for a midsize data center (around 2,500 square feet); it can cost more than $200,000 a year just to maintain an acceptable temperature.
Key Data Center Opportunities
- Advances in cooling technologies save money and reduce energy consumption.
- Rationalized IT systems make it easier to respond to changing demands.
- Reclaimed floor space and lower energy bills help realize the full benefits of IT consolidation efforts.
- Reduced complexity facilitates maintenance and allows more flexibility.
- Energy-efficient, optimized data centers prepare an organization for the data requirements of tomorrow.
Multiple Points of View
Several forces prevent organizations from dealing with data centers issues. One of the biggest challenges of evaluating data center options is getting all of the different points of view involved to converge into one coherent strategy. Different departments view data centers in different ways. The IT department looks at them in terms of the servers, storage devices and switches. The facilities department, on the other hand, sees them as physical structures with structural, electrical and plumbing challenges. Meanwhile, CFOs see them in terms of capital expenditures, ROI and budgetary constraints.
Although it obviously involves information technology, developing a data center strategy requires an engineering orientation that not all IT departments or traditional systems integrators have. Taking a holistic approach is the key to successful planning and execution. For example, a plan may require specialists with design, integration and implementation of telecommunications systems experience and their related infrastructure components.
Data center development requires a diverse range of skills:
- Data center design and full-service planning
- Data center relocation
- Site selection and preparation
- Asset inventory, tagging and move planning
- Asset moving and installation
- Disaster recovery/redundant site setup
- Project management
- System and data migration
- System/server consolidation
- Cabling, power and physical infrastructure services
- Post-migration maintenance and support
A phased approach is often the best option -- beginning with a period of discovery and assessment that consists of a review of the existing and planned data center space requirements and overall IT infrastructure. (See data center best practices at the end of the article.)
Technical requirements that need to be detailed include:
- Power and cooling densities, as well as loads by equipment type and areas
- Space requirements
- Physical parameters (reliability, power, cooling, fire protection)
- Disaster recovery, backup strategies and equipment
- Telecommunications and connectivity requirements
- Networking infrastructure
- Security (physical and network)
- Operations and staffing space
- Equipment inventory (server and storage)
Organizations that set off on their own to develop a data center plan often go first to a firm that specializes in mechanical, electrical and plumbing - where they are met by a perfectly capable electrical or mechanical engineer who knows very little about IT. The result is a facility that doesn’t quite fit IT - not unlike what happens when an application is developed without the participation of the users in the process.
It’s important to work with a team of experts who have experience with the larger context for all of the specialties that go into a cohesive data center development. Once all of the appropriate information has been collected and prioritized by stakeholders, there are three basic options for data centers: in-house private build, co-located and outsourced.
Each of the three alternatives have pros and cons that need to be evaluated in order to determine the best data center option or combination of options. It often depends on how an organization likes to handle capital expenditures. Organizations that like capex on their books and are looking for an ROI of between five and 10 years tend to favor building their own data centers. The larger the organization, the more difficult it is to make the economics of co-location work. The economics of outsourcing make it a better option for small or midsized companies with standard platforms. The cost of outsourcing escalates rapidly with size and complexity and can be very expensive to take back control from the outsourcer.
It’s important to make the distinction between outsourcing and managed services. Managed services is not outsourcing. It’s similar to co-managing and can co-exist with either an in-house or co-lo data center. A managed service offering can take on all or some of the management of a client’s IT, so they get some of the features of outsourcing, but for a much better price. And the customer stays in control.
One of the most potentially contentious discussions that must take place in the development of a data center strategy has to do with determining acceptable levels of redundancy. There are four tiers of redundancy, the highest of which is called Tier 4.
Tier 4 redundancy ensures “concurrent maintainability,” which means that everything works all of the time and that any piece of supporting equipment can be taken
offline without affecting the data center. It sounds great, but it involves having two of everything in the data center in case one of anything goes down. Everyone wants that until they see the price. Deciding which systems need full redundancy versus which can tolerate downtime depends on risk management parameters that vary with different subsystems and should be negotiated internally. Most organizations start out wanting Tier 4 but end up acknowledging that some downtime, for at least some systems, is acceptable.
Appropriate tiering levels need to be accommodated in the design of a data center as well as any plans for off-site disaster recovery locations. Having to retrofit critical subsystems for redundancy is one of the reasons that many existing data centers have become so crowded.
Migrating systems to a new or refurbished data center has the potential for as much high drama as a brain transplant. Project management skills are critical - it’s not like moving a desk. Redundant equipment needs to be functioning for those systems that cannot be interrupted. A new network infrastructure is recommended to ensure optimal connectivity. The actual migration of devices needs to be planned in stages, and the plan needs to allow for some potentially unpleasant surprises. A trial run in which select systems are shut down before they are moved so an organization knows ahead of time where the issues are going to be can save a lot of headaches. The goal of an uneventful migration is accomplished only by anticipating anything that could go wrong and then making sure it doesn’t.
Developing and implementing a data center strategy is an admittedly long and involved process, which for large complex data centers can take a year to complete. The best data centers are dynamic systems that can evolve as data demands change. Besides producing a state-of-the-art data center, the discipline of having identified all aspects of what it takes to keep data flowing smoothly results in a beneficial degree of awareness and the establishment of lines of communication.
10 Data Center Best Practices
Bring together all of the stakeholders early in the process. Developing and implementing a successful data center strategy depends on collaboration not only between IT and the facilities department, but also the CFO’s office.
Understand that specialists can do only what they know. Don’t ask your IT team about cooling systems or expect an MEP (mechanical, electrical, and plumbing) engineering firm to understand IT-specific needs.
Measure the power demands of different types of components in your data center and establish benchmarks and measurable standards to aid in ongoing power management. Managing power effectively requires knowing all the variables.
4. Go Green
Advances in cooling design concepts, such as high-density cooling and controlled heat rejection, can save significant expense and reduce the carbon footprint at the same time. For example, in cooler climates, an organization can take advantage of ambient outside air and let Mother Nature help the data center stay cool in the winter.
Decide what requirements the data center needs to meet and choose a small number of technologies and vendors to meet them. Lower technical complexity translates into lower operational overhead. Keep it simple.
Virtualization makes it possible to reduce the number of physical servers needed by an average of 10 to 1 and ensures that all of your servers are running at appropriate load levels. Run smart.
Make sure all of the data center systems are performing at their best. New applications and communications architectures, such as software as a service (SaaS) and unified communications, require that everything in your data center not only works well by itself, but also works well together with everything else.
Read and internalize EIA/TIA-942, the compilation of guidelines for everything about data centers. The more standardized you make your data center, the easier it will be to maintain and upgrade -avoiding having to retrofit later.
Automating routine processes can reduce operational cost, free up valuable human resources, and standardize procedures. Take advantage of the Information Technology Information Library (ITIL) change management guidelines.
The data center, like the data contained within it, is a corporate asset. Designing, implementing, maintaining and paying for it should all be done in line with the overall corporate strategy. The data center doesn’t belong to the IT department or facilities department. It belongs to the corporation at large.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access