HP P2000 G3 MSA Technical White Paper page 18

Technical white paper | HP P2000 G3 MSA
In both caching strategies, active-active failover of the controllers is enabled.
You can enable and disable the write-back cache for each volume. By default, volume write-back cache is enabled. Data is
not lost if the system loses power because controller cache is backed by super capacitor technology.
For most applications this is the correct setting, but because backend bandwidth is used to mirror cache, if you are writing
large chunks of sequential data (as would be done in video editing, telemetry acquisition, or data logging) write-through
cache has much better performance. Therefore, you might want to experiment with disabling the write-back cache. You
might see large performance gains (as much as 70 percent) if you are writing data under the following circumstances:
Sequential writes
•
Large I/Os in relation to the chunk size
•
Deep queue depth
•
If you are doing any type of random access to this volume, leave the write-back cache enabled.
Caution
Write-back cache should only be disabled if you fully understand how your operating system, application, and HBA (SAS)
move data. You might hinder your storage system's performance if used incorrectly.
Auto-write through trigger and behavior settings
You can set the trigger conditions that cause the controller to change the cache policy from write-back to write-through.
While in write-through mode, system performance might be decreased.
A default setting makes the system revert to write-back mode when the trigger condition clears.
To make sure that this occurs and that the system doesn't operate in write-through mode longer than necessary, make sure
you check the setting in HP SMU or the CLI.
You can specify actions for the system to take when write-through caching is triggered:
Revert when Trigger Condition Clears: Switches back to write-back caching after the trigger condition is cleared. The
•
default and best practice is Enabled.
Notify Other Controller: In a dual-controller configuration, the partner controller is notified that the trigger condition is
•
met. The default is Disabled.
Cache configuration summary
The following guidelines list the general best practices. When configuring cache:
For a fault-tolerant configuration, use the write-back cache policy, instead of the write-through cache policy.
•
For applications that access both sequential and random data, use the standard optimization mode, which sets the cache
•
block size to 32 KB. For example, use this mode for transaction-based and database update applications that write small
files in random order.
For applications that access only sequential data and that require extremely low latency, use the super-sequential
•
optimization mode, which sets the cache block size to 128 KB. For example, use this mode for video playback and
multimedia post-production video- and audio-editing applications that read and write large files in sequential order.
Parameter settings for performance optimization
You can configure your storage system to optimize performance for your specific application by setting the parameters as
shown in the following table. This section provides a basic starting point for fine-tuning your system, which should be done
during performance baseline modeling.
18