Network Computing is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Capacity Planning: View From the Trenches

It's difficult -- if not impossible -- to generalize about storage capacity planning. Talk to a storage customer, and it's evident there are at least as many approaches as there are storage setups. Talk to a vendor, and prepare to have the parameters of the conversation blown out to include the alpha-to-omega of data management -- including policy networking, tiered storage, and lifecycle management.

A few basic tenets apply: The old mainframe-era adage of maintaining 50 percent to 60 percent capacity thresholds remains the conventional wisdom with SANs and other networked storage. But some other technical and business realities have emerged to complicate things -- namely, the advent of virtualization, increased IOPS (input/output operations per second) demand, and the ability to tie storage consumption to business requirements or strategic initiatives.

In an effort to discern some common best practices, Byte and Switch turned to those in the trenches for input. Following are comments from storage managers about their most efficient means of predicting storage requirements for IT budget planning.

Use multiple measurements

Excel spreadsheets are perfectly adequate for online media site Alibris when it comes to capacity planning, says CTO Michael Schaffer. It's what you put into the spreadsheet that counts. And Schaffer is interested in two major metrics: raw storage capacity and performance (IOPS and throughput).

"Our business drives growth in the demand for these services. We have a multiyear financial plan (in dollars), which we map to order volume and inventory stats, like number of media vendors in our network and the number of SKUs in our catalog," Schaffer says. "The vendor count, SKU count, and order count all drive our expected storage needs, in Gbytes and IOPS."

  • 1