Assessing storage efficiency
Metrics, assessment process
By Kevin McIsaac
October 15, 2002
To avoid "analysis paralysis" and lengthy benchmarking, ITOs should start
with a small and simple set of metrics, such as the following:
· Managed storage per administrator: This includes all time spent by staff on
all activities in all life-cycle phases required to manage the storage (e.g.,
deployment, upgrade, provisioning, monitoring, and backup/recovery). The current
benchmark for distributed systems is approximately 1.5TB per administrator. ITOs
should aim to increase this ratio in sync with their projected storage growth
· Cost of storage: This benchmark varies widely depending on the class of
storage and decreases over time as storage hardware prices decline. This metric
must include all storage related hardware, software, and maintenance costs.
· Percentage utilization: This is the percentage of installed storage that is
utilized. Distributed systems, with direct-attached storage, can be as low as 30
to 40 percent for small NT servers and 40 to 50 percent for larger Unix servers.
Projects (such as storage consolidation) that increase this to more than 70
percent can have a significant impact on the TCO.
Because these metrics are highly dependent on factors such as workload (database
versus file and print) or storage SLAs, they should be calculated independently
for significantly different of elements of the storage portfolio (e.g., by
machine type--mainframe, Unix, or NT--by storage SLA).
Storage assessment process
To calculate these efficiency metrics, ITOs should follow a simple storage
assessment process as follows:
· Step 1: Develop collection forms by asking a range of staff members for a list
of the significant storage tasks undertaken by the storage group. Although this
list should be comprehensive, ITOs should avoid overcomplicating collection by
adding too many items. A second form is required for listing all the major
systems, the total storage capacity, and the used capacity.
· Step 2: Collect data by circulating the activity list to all operations staff
involved in storage operations. Each member completes this list by recording the
number of hours he or she spends on each activity in a typical month. Storage or
systems administrators are accountable for recording the total storage capacity
and the used capacity used for each system.
· Step 3: Analyze results by calculating the metric and reviewing the activities
for the largest time sinks (usually backup and recovery) to be targeted in the
· Step 4: Transform storage operation by defining and initiating projects that
affect these metrics. Typical projects are consolidation and automation of
backup and recovery and developing an enterprise storage network. Other projects
include process refinement, role consolidation, and staff reduction.
Steps 2 through 4 must be repeated on a regular basis (e.g., six months) to
monitor the effect of the transformation projects. Although this process can be
performed on most infrastructure components, storage is usually the
highest-impact starting point.
Although ROI is appropriate for business projects, infrastructure projects are
better evaluated on simpler and more direct cost efficiency models.
To constrain a rapidly growing storage budget, ITOs should conduct a storage
assessment process that focuses on a few concrete efficiency metrics and then
implement projects that transform storage operations to improve these metrics.
SOS: Save on Storage
First published October 3, 2002
By Kevin McIsaac