Better capacity forecasting

There are two methods for devising storage capacity forecasts: quantitative and qualitative. By combining the two, you can develop practical metrics that will make more accurate forecasts.

Just-in-time manufacturers like Dell Inc. know what they need when they need it. This lets the company carry a...

smaller inventory, lower costs and still meet demand for its products. Many manufacturers have moved to streamline their forecasting processes to better anticipate and plan for customer demand. Storage managers can realize similar benefits and avoid overbuying storage by developing a forecasting methodology and creating the metrics to track it.

There are two basic approaches to forecasting: quantitative and qualitative. With storage resource forecasting, it's good practice to use both methods. In quantitative forecasting, statistical analysis is used to analyze historical resource consumption to provide a basis for determining future resource needs. Quantitative analysis provides a predicted growth rate for storage resource consumption and gives some insight into seasonal peaks that may occur, such as yearly application rollouts or semi-annual migrations to data warehouses. The chief drawback of using only quantitative forecasting is that history isn't always an accurate predictor of the future. Qualitative forecasting helps by adding the real-world insight that's required for an accurate forecast.

Once the quantitative method has been completed and a base forecast is established, use the qualitative method to complete the forecasting process. While the quantitative approach provides historical information, the qualitative method anticipates future needs based on planned activities. With qualitative forecasting, information is gathered directly from users through regularly scheduled resource-planning meetings. During these meetings, users should provide information regarding new initiatives that may require more or less storage.

Forecasting metrics

While using the quantitative and qualitative methods provides great insight into developing a forecast, it's important to be flexible. There are always outside influences and pressures--such as regulatory compliance, mergers and acquisitions, and general economic cycles--that can greatly impact storage resource needs. The following tips will help you develop an accurate forecasting process:

  • Start at the top. Communicate the value of forecasting to senior IT management and the business units. Ask them to sponsor the forecasting initiative; key stakeholders are more likely to actively participate if management endorses the effort.
  • Develop accurate metrics. It's crucial to develop the appropriate forecasting metrics so they're meaningful to all interested parties. Once a forecasting initiative has gained the attention of senior management, meaningful metrics will solidify the value of the initiative for senior managers. Uncertainty about the accuracy of the metrics can undermine credibility and seriously damage a forecasting initiative. It's important to ensure that storage reporting metrics can be generated on a regular basis and without a great deal of effort, so try to automate the process of collecting data and generating reports.
  • Talk to vendors about capacity on demand. A low-risk way to build flexibility into a resource plan is to negotiate a capacity-on-demand buying process. Under such a plan, a vendor installs the maximum capacity gear, but pays only for the capacity used. Capacity on demand makes it easier and faster to add storage capacity by eliminating or reducing setup and configuration time. Vendors are amenable to capacity-on-demand contracts because they improve their chances for additional sales.
  • Incorporate compliance policies. Corporate compliance policies can affect the amount of storage capacity required. Organizations often save information in multiple forms (electronically and on paper), resulting in wasted storage capacity. Corporate compliance policies can address this by defining the types of information to be saved and where it should go.
  • Integrate storage with a single IT capacity-planning initiative. A storage forecasting process should be integrated into a general IT initiative chartered to perform capacity planning for all IT resources, including storage, servers, networking, telecommunications, facilities, headcount and so on. It's important to leverage the many activities that contribute to the capacity-planning process, which includes forecasting, budgeting and general business planning. This kind of initiative is usually driven by the CIO or another high-level executive within IT.

Capacity discovery tools
You can't develop an accurate capacity forecast unless you deploy a storage resource management (SRM) tool. SRM agents poll hosts, servers and storage network switches to gain insight into storage capacity, identify unused capacity and track storage usage patterns over time. Storage admins often request additional capacity without understanding how current storage capacity is allocated to support a server or application. Allocated capacity can be viewed at three levels: LUN, file system and database. All three levels should be analyzed and considered when determining capacity needs.

Benefits of storage resource forecasting
Better predictability and manageability. A forecasting process shifts the burden of predicting future capacity needs from the storage administrator to the business units.

Ensures the availability of storage resources. By aligning more closely with users to understand future capacity needs, storage managers can better plan for that demand and ensure the necessary resources are available.

Increases productivity. With a managed forecast, there's less firefighting. Storage managers can justify new storage purchases far in advance and work in a more strategic mode to determine better ways to manage their resources.

Decreases costs. Vendors are more apt to provide greater discounts for larger dollar deals. By limiting the number of purchases, storage administrators can reduce costs by taking advantage of favorable purchasing periods, such as the end of a quarter when vendors are under revenue pressures.

Identifying unauthorized files is a touchy subject for most organizations, but many employees save personal files like music, movies and other non-business-related data on corporate resources. These unauthorized files can consume hundreds of gigabytes--even terabytes--of corporate resources. SRM tools can quickly locate these files by their meta data. An SRM program can also identify files that haven't been used for long periods of time and, according to preset policies, delete them or migrate them to a lower tier of storage.

Turning on an array's oversubscription feature is another way to find unused capacity. For example, with the oversubscription feature turned on, if a user asks for 1TB but uses only 500GB, the storage array can free up 500GB. It's important, however, to closely monitor actual utilization when this feature is turned on to avoid running out of capacity on a particular array. Virtualization products also increase overall capacity by allocating storage across multiple arrays into a single, large pool of storage that improves the capacity utilization in each array.

The ultimate goal
There are many benefits to implementing a storage forecasting process, but the ultimate goal is to change user behavior. Once a storage forecasting process is in place and functioning properly, storage managers can introduce a reward system for users and business units. Those users who forecast their demand accurately and utilize their resources wisely will achieve substantial cost savings. Developing a comprehensive forecasting process is a critical step in moving to a mature, predictable and controllable storage management environment.

Dig Deeper on Storage management and analytics