14 Adaptive Optimization - HP 3PAR StoreServ 7200 2-node Administrator's Manual

14 Adaptive Optimization

Adaptive Optimization (AO) uses System Reporter statistics gathered from LDs and physical disks
to relocate customer data on physical volumes in an optimal way. AO relocation accomplishes
two primary goals:
Increase performance of frequently accessed regions of data by moving those regions to
higher-tier storage (for example, moving to SSDs from normal spinning media).
Improve cost-efficiency by moving lightly accessed regions of data to a lower performance
and less expensive tier of storage (for example, moving from regular drives to NL drives).
Because storage tiers can be of different RAID types, capacity efficiency is maximized by using
only RAID1 for the most frequently accessed storage and by using RAID5 or RAID6 for less frequently
accessed storage. Other benefits include:
AO can migrate data from a fully occupied tier of storage to another tier that has more
available capacity.
AO can also be regularly scheduled so that it can adjust the data layout as your data usage
changes over time.
AO has been integrated into the HP 3PAR OS as part of changes in the 3.1.2 release. A simple
user interface has been developed whereby AO can be started either as a scheduled task or as
a manual command using the new startao command. For more information about tasks, see
"Managing Tasks" (page
NOTE: Active use of AO requires an Adaptive Optimization license. Contact your local HP
representative for information.
AO is built on top of the new version of System Reporter (SR), which also now runs as part of the
HP 3PAR OS.
SR must have been actively gathering data on VV regions for a period of time (a minimum of 3
hours, but preferable for a longer time including any time periods where the storage system has
had peak load/usage).
The startao command is then used to perform an analysis of the data collected by SR to identify
regions within virtual volumes that are either heavily used or lightly used, and then generates a
series of secondary tasks to move these regions to faster or slower storage tiers.
Running AO on the HP 3PAR OS itself offers these advantages:
AO configurations can now be created, modified, and removed using the HP 3PAR CLI or the
HP 3PAR MC.
The database scheme has been restructured on node to be more efficient and reliable.
The actual movement of data can use data from a given time period in the past rather than
only from the immediate past. This means data movement can occur at low-utilization time
periods while using an analysis of statistics gathered during peak periods.
A time limit can be set for data movement so that scheduled data is moved only during
low-utilization periods rather than during peak periods.
An AO configuration is specified as part of the startao command. When this is run as a scheduled
task, details of the analysis and generated move commands can be viewed using showtask
<task id>, where <task id> is the ID of the startao command.
140 Adaptive Optimization
126).
d