Data Center Decisions
The advent of new computing technologies and expectations for 24 x 7 uptime have posed a stiff test for data centers. Rack-mounted high-density clustered servers and blade servers need more power and generate more heat than the servers of the past. And as data centers age and more equipment is added, current electrical, HVAC, uninterruptible power supplies (UPS) and cable management systems become over-burdened or even obsolete. The bottom line is that many existing data centers are not currently equipped to handle the new electrical and cooling demands of modern IT infrastructure. Addressing the problem could mean a retrofit, complete overhaul or relocation of the data center.
While the prospect of a new data center is exciting, it is a major undertaking that requires the cooperation and collaboration of two key groups in the organization — facility management and IT. These two groups can be so fundamentally different from each other that they speak different languages, and each has its own agenda and priorities. It’s no small task to get them going in the same direction.
Let’s take a look at the two groups. Facility executives know every nook and cranny of the buildings that house the organization. They run on a tight budget, and their work is highly visible to the CEO and CFO. IT workers, on the other hand, manipulate bits and bytes to make sure that data keep moving, computers and networks are up and running, and intangible assets of the business stay safe and secure. To upper management, IT is a black box. Executives have to have a certain amount of blind faith to entrust their business assets to IT. Consequently, IT professionals have a great deal of power without as much financial accountability.
When the time comes to consider a data center retrofit or relocation, IT usually makes the request. Facility management professionals often don’t have the expertise to do proper discovery, yet they are expected to handle the execution. IT professionals are trained in daily operations. Chances are, they do not have a solid understanding of the construction process and are not trained to do proper discovery. Often there is a breakdown in communication between the two groups. This is where an objective third-party consulting group can help. Outsourcing the IT discovery, design, planning and project management services for the technology infrastructure can help to bridge the gap between facility management and IT, and facilitate a faster, smoother data center retrofit or relocation.
Data centers built today look vastly different from those of even three to five years ago. Today, most companies have a series of clustered servers chained together. There may be as many as 40 per rack, consuming power and generating heat at 10 to 15 times the rate of the servers they replace.
Some large corporations have upwards of 10,000 servers. As a result, the electrical and cooling requirements are much more challenging to meet. The Uptime Institute estimates that 70 percent of the construction cost of a new data center is allocated to electrical and mechanical systems, compared with only 15 percent in an office building. And, as uptime expectations continue to grow, the performance demand on data center electrical and HVAC systems also increases.
The Uptime Institute notes three primary drivers for either retrofitting or building a data center:
- Existing power or cooling capacity is too small
- The tier of functionality (see sidebar)
- The amount of computer floor space
In evaluating whether a data center is adequate, it’s important to look at business plans. Perhaps the business is in expansion mode or in the process of migrating legacy equipment to newer technologies that will enhance business operations and efficiency. If more servers or computer labs are going to be added, for example, facility management will need to start making plans for housing the equipment, which may mean a remodel or relocation to a new facility. Additional electrical and mechanical capacity will be necessary to support the new devices and technologies.
Time and financial considerations will also play heavily into the decision about whether to retrofit or build from the ground up. A new building may cost up to four times as much as a retrofit, yet the cost can be fully justified by the demands of business growth projections. It takes about two years to design and build a new data center and approximately six to 12 months to retrofit an existing one.
Building a new data center may be more logical than upgrading an existing facility to accommodate requirements of new hardware and technologies. An assessment should include risk factors that are associated with a retrofit project. A retrofit will inevitably require some planned downtime, but there is always the risk of unplanned downtime. Some organizations will justify building a new data center rather than risk an unplanned outage.
No matter how small or how large the data center, a thorough data center assessment is a critical first step. Developing a realistic budget and schedule requires the cooperation of IT and facilities management to set expectations and understand the depth and breadth of the undertaking. Generally, if this is handled by a consultant, it takes about four to ten weeks and can cost an organization anywhere from a few thousand dollars to tens of thousands, depending on the size of the data center. It is important to come out of this step with a high-level report detailing the findings in language and drawings that can be understood by all involved parties. This initial assessment saves a lot of potentially costly and time-consuming missteps later.
After gaining approval and consensus on the parameters of the data center project, execution can begin. This involves collaboration with all the team members to make sure the project goes smoothly every step of the way.
Streamlining the Process
A data center retrofit or relocation project can be broken down into three phases.
Phase 1 involves discovery, scheduling, developing budgets and assembling the team.
Identify the team. Project team members include internal staff, department heads, budget approvers, and external partners and vendors.
Define the objectives. What is to be built or retrofitted? List all of the technology rooms needed. In addition to the data center, consider the minimum point of entry (MPOE), main distribution frame room (MDF), server rooms, network operations centers (NOC), wiring closets or intermediate distribution frame rooms (IDF), technology labs, executive conference centers, theaters or presentation rooms, security rooms or wireless rooms.
Establish a method of communication between IT and construction teams. This usually means weekly meetings to hash over IT budgets and brainstorm on IT and facility requirements.
Gather information and set expectations. The burden is on the team to agree on the level of redundancy required. This will ultimately determine the budget and schedule. Create an equipment list to determine load requirements for power, cooling and space. Develop preliminary rack layouts and develop multiple options and budgets to help in the decision process.
Get executive sign-off on priorities and budgets. Business and financial issues must be discussed with upper management and business priorities should be established. For example, develop a summary report that explains the reliability tier concept — the various options that are available and the associated costs. More redundancy will usually be more expensive and take longer. Teach senior management about the trade-offs and get an informed directive before the project begins, so that everybody is on the same page.
Draw up a preliminary schedule. It’s critical to fix a realistic move date; establish target completion dates for key milestones such as room buildouts, installation of electrical, cooling, and cabling systems; set cable testing procedures and mechanical commissioning procedures; schedule dates for testing data circuits and phone lines; and, finally, set the date when the new facility will be occupied and the old facility will be de-activated.
The second phase entails construction project management.
Create detailed IT bid documents and construction drawings. Cover items such as documentation and schematics, rack layouts, cable management, electrical design and capacity requirements, cooling, air flow requirements, floor plan layouts for each IT room, fire suppression, seismic requirements and cabling specs.
Communicate with vendors during construction. To ensure that important information reaches all parties there should be regular design and construction meetings for vendors and weekly meetings with IT and facilities. They help negotiate contracts with vendors and ensure that drawings are updated, change orders are tracked, and IT circuits and network connections are tested.
The third phase is relocation project management.
Create an IT move plan. This phase involves setting priorities, placing orders for data and phone services, tagging each piece of equipment, lining up insurance, making sure systems are shut down prior to the move, and seeing to it that high-priority systems are moved first. After the move is successfully completed, training should be provided for users on the new systems.
Post-move audit. An assessment with vendors and stakeholders will help tie up loose ends. Lessons learned should be documented and implemented in future projects.
Develop policies and procedures for IT. In this final step, complete documentation should be developed that details policies and procedures for all new systems. These include as-built drawings of electrical, cooling, cabling, fire suppression and security so that IT can function smoothly going forward. What happens when the alarms are activated, for example?
The biggest mistake people often make is to assume that this is a good time to upgrade the network or software. But it’s not a good idea to make changes during a move because it can complicate and delay the process. It’s best to make major changes before or after, when the dust settles and IT can focus on what it does best — optimizing software and the network performance.
Levels of Data Center Reliability...
The Uptime Institute categorizes businesses into four tiers:
...Determine Facility Infrastructure Needs and Costs
Retrofit or Build New?
Gregory Bell is director of client services for Teladata, an IT infrastructure consulting firm. He is responsible for project management services, strategic planning, and client relations. He also has overall responsibility for project design, infrastructure budgets, schedules and project costing.