In my white paper on Storage Economics, we discuss the several activities or initiatives that IT departments can undertake to reduce storage OPEX. These activities fall into a range of effort and paybacks, as depicted in this graphic.

HDS consultants have documented the top 6 activities to reduce storage OPEX; this entry will deal with the first, storage consolidation.

The cost of storage ownership can be less if there are fewer things to manage, control, and operate. Aggregating storage costs from many smaller arrays or DAS environments is more costly than a larger, centralized architecture. Where geographic consolidation is practical, OPEX costs can be reduced in the areas of:

• Hardware maintenance
• Hardware and software depreciation expense or lease payments
• Storage-based software maintenance
• Front-end and total SAN ports (whatever the protocol)
• Raised floor space in the data center
• Cost of electricity and air conditioning
• Cost of waste with stranded capacity (stranded LUNs)
• Labor and management efforts can be consistently applied, usually with common tools
• Possible implications of centralized storage management labor and organizations
• Larger, enterprise arrays are engineered for higher availability when compared to the aggregate utilization rate of smaller arrays. This availability difference can be converted into a cost to the organization.

This effort is closely interrelated with other activities such as SAN and best practices; these will be discussed in future lessons.

Now to do a side-by-side comparison of smaller disaggregated storage arrays vs. single/fewer larger arrays, use the cost areas listed above, and create cash flows for each line over the next 3-4 years. Do the same with (a possibly new) larger array acquisition. Cost savings can be compared to the capital acquisition to develop the ROI, NPV, IRR and payback time of such a purchase.

This recommendation does not support the notion of international consolidation for large enterprises. The cost of long distance circuits and the performance problems associated with latency typically do not offset the resulting OPEX factors. Local consolidation can be within the datacenter, buildings or metro campus area where high-speed FC connections are possible (10km radius).

Comments RSS