By Mark Bortle

In the midst of today’s world of rapid change and rapid response, data center managers must decide where to spend their time. Capacity planning and capacity management usually end up at the bottom of their lists. However, today capacity management is becoming more critical. When an enterprise experiences a merger, an acquisition, a budget cut or a need to respond to a competitive threat with enhanced technology, C-level executives start asking data center managers tough questions about costs and capacity. For example:

 • Can our existing data centers absorb the IT assets of the company we are about to acquire or merge with?
• Can we consolidate into these data centers to reduce our total IT spend and optimize total cost of ownership (TCO)?
• When will we have to expand our data center footprint at current projections?
• How long will it take me to provision for growth?
• Do we have a predetermined capacity point at which we must consider data center expansion, procurement or managed service options? How long will that take? What is this dependent upon?

Why not get ahead of the curve? Data center capacity management is the management discipline of tracking and accounting for the use and projected use of data center resources including power, cooling and space. This management discipline can greatly facilitate planning and decision making activities associated with business strategy and growth.

Power consumption and cooling requirements are top expenses of any data center. How hardware assets are deployed affect the consumption of expensive data center floor space. Producing a detailed data center capacity management plan provides the framework and policy necessary to oversee the allocation and subsequent use of these precious resources. It will also aid in lowering TCO over the projected service life of the data center.

As the global economy grows and expands on the backs of technology innovation, one of the toughest issues facing data center managers is how to handle additional growth due to power, cooling or space constraints. Insufficient data center capacity can prevent growth right at the moment when an organization needs it the most in order to take advantage of market changes. The build out and provisioning of a new data center can take 18–24 months. Data center capacity management planning puts a framework around pro-actively managing data center power, cooling and space as critical, finite resources necessary to support the organization and sustain projected growth. With a detailed capacity management plan, data center managers can efficiently allocate required power, cooling and space based upon the physical (usable floor space, provisioned power distribution, installed cooling) constraints of the data center…and rest easier when executives come knocking.

It is important to gather and perform analysis on key data center specific data to answer critical questions. This analysis will provide content for the data center capacity management plan and upward communications to company executives. The content may be expressed in text or combined in useful graphs as part of a dashboard.

Best Practices
What are the key components of a data center capacity management function? At a minimum, they involve a plan, a process, data gathering, analysis, corporate’s blessing and communication.

• A Plan—a structured, executable document that outlines the corporate program, the data center capacity management continuous process, and addresses how data is collected and analyzed to produce capacity information for the following:

  Physical aspects for each data center within scope
    Data center construction and size, square feet of usable, admin and other space
    Geographical data

  Power distribution for each data center within scope
    Power distribution feed locations
    Metered points of entry
    Power capacity
    Power consumption
    Authorized capacity (from external power providers)
    Time to provision additional capacity if required

  Cooling distribution for each data center within scope
    Cooling system type and capacity
    Cooling system consumption
    Cooling system upgrade or change considerations
    Time to provision additional cooling

• A Process—Do not just create a plan and leave it in a drawer never to be updated or changed as the enterprise changes. Create a well-designed and documented continuous improvement process for managing capacity. Assign process champions or owners who maintain current revisions and updates as required. Seek to automate as much as possible and leverage industry toolsets where possible.

• Data Gathering—Create a data gathering process that is reviewed and generally accepted by all stakeholders and decision makers. Make sure this process is identified, documented and communicated well to ensure that the most meaningful and accurate data is collected and analyzed for capacity planning purposes. Data drives the credibility of the effort and falls in line with the all too familiar garbage-in, garbage-out concept.

• Analyze—Coupled with gathering the correct data, the analysis method should also be generally accepted within the organization. Calculations for current power capacity, current power consumption, planned power consumption and projected capacity levels should be reviewed and agreed to by key stakeholders. It is critical for all stakeholders to have confidence in the results of the analysis.

• Corporate Blessing—Often missed, this is a critical step to creating a successful plan. Obtain the support of corporate leadership, one which is supported by corporate policy and formalized communications.

• Communicate—If the data is collected but not analyzed, reported or assimilated by the right viewers, then the entire exercise of data center capacity management is a waste of time. Publish the information obtained through capacity planning to the right people so they can make good decisions. For organizations that already have established performance metric reporting, data center capacity management findings can and should be combined into the normal reporting stream in order to efficiently communicate with C-level executives.

It is important to summarize that the data center capacity management function is similar to any new business process. If it is recognized and supported by corporate leadership…if it is a formalized, continuous process supported and defined by a structured plan…if the data and data analysis that supports the plan is socialized and widely accepted by key decision makers and stakeholders…if executives are aware of the importance of data center capacity management and regularly review the capacity metrics…then the program will be successful.

Critical Steps to Success
The biggest issue with a data center capacity plan is to gather and analyze the data to produce accurate and meaningful information. Some key steps to minimize this effort are:

  1) Design and validate a sound data center floor plan. This will take into consideration the type of rack, its space capacity in U’s, and the total kW configured for the rack. Ensure that hot isle/cold isle and alignment with floor tiles is considered. Provide enough space between rows and large enough cabinets to facilitate hardware maintenance and structured cable routing (and placement), respectively. Combining this effort, determine rack density and space capacity.

  2) Determine power distribution specifics. Identify power distribution routing to UPS, PDUs and generator (capacity, time) feeds. Identify power configured to each rack and circuit configuration (AMPs and kW to rack). Is each administrative space configured into circuit panels for data center distribution? If so, what percent is that of total?

  3) Determine cooling capacity and air flow. Identify and track the key metrics of cooling systems installed. Is the cooling system modular? Can its capacity be increased with upgrades easily? What is the BTU/hr rating? Is airflow sufficient for the density planned in the data center? Consider leveraging computation fluid dynamics concepts and software to model air flow within the data center.

  4) Know your assets. Integrate the data center capacity management process into the organization’s configuration management database (CMDB) to track capacity management specifics. Power consumption, cooling requirements and rack (U) size are key metrics to track for hardware assets configured in the data center. Cooling may be converted from BTU to kW to keep metrics in the same unit of measurement. Also, track non-standard hardware assets such as storage arrays, tape silos, mainframes and associated peripherals. This is where standardization also pays off. Standardize on energy-efficient hardware that meets the objectives of the organization. This effort will also contribute to easier management, data collection and analysis.

  5) Integrate capacity management with demand management and hardware deployment. Business units will produce demands for IT infrastructure. Determine the data center capacity management metrics required in the requirements specification to accommodate and track the desired hardware assets. Specifically, hardware type, model, rack size (U’s), power configuration (power supplies), max power rating (kW) and cooling requirements (BTU).

  6) Evaluate automation toolsets to support data center capacity management. Toolsets may be a combination of hardware and software. Power strips configured into the racks can be managed by software. Power strips can be purchased with LED power indicators to show the power draw per strip. Some software can be leveraged to provide a visual representation of the data center and model air flow as well. Toolsets are on the market today that can provide monitoring and reporting of data center capacity. Other toolsets can be leveraged for asset discovery and populating a CMDB.

Many data center managers support this effort by using informal processes and spreadsheets populated with questionable data. This leads to an inaccurate end product as a result of manual process intervention. In some cases, data center resources are assumed to be abundant based upon day-to-day operations: We have unlimited power…we have plenty of space. Either of these attitudes can lead to serious issues when an organization announces a merger, acquisition or growth forecast, which brings increased demand for data center resources. C-suites do not want to hear that the organization does not know its data center capacity, utilization, time to provision additional capacity, or that they have to invest $50 million and 24 months in a new data center build or sign a costly contract for managed services for two to five years. A better scenario is to incorporate the practices mentioned above, plan ahead and look like a hero.

Mark W. Bortle, is a senior manager in the data center and infrastructure practice of Acumen Solutions, Inc., a business and technology consulting firm. He can be reached at mbortle@acumensolutions.com.