User Manual

Appendix
B-7
algorithms that can intelligently self-monitor and self-adjust the
performance parameters.
• Adaptive read-ahead (pre-read, pre-fetch) optimization
Read-ahead operation improves the performance of sequential reads by
pre-fetching data from disk drives according to current hosts’ read
commands. The firmware can further identify multiple sequential read
streams in random access and perform pre-read for the streams. The
administrator can also specify the pre-read depth for dynamical tuning.
• Configurable write caching policies
Write cache can improve the response time and concurrency level of hosts’
write commands. With the write cache, the controller can merge
consecutive write commands to single write command and lower the disk
drive’s utilization by avoiding over-write commands. On the other hand, to
ensure the best data reliability, the write policy can be set as write-through
to make sure all data is written to the disk media.
• Performance monitoring
The controller keeps extensive IO statistics for performance monitoring. The
statistics include physical components, like host ports and disk drives, as well
as logical objects, like LUN, cache, and logical disks. The complete picture
of the storage performance profile is presented and performance tuning
can be conducted more effectively.
• Intelligent IO processing
Intelligent IO processing algorithms are efficiently executed to optimize the
command execution and streamline the data flow. Disk IO scheduler is
deployed to reduce the number of disk access and minimize the seek time
among disk access. Elaborated RAID algorithms are performed to minimize
the number of parity update and shorten the response time. The
administrator is allowed to control these optimizations and tune the
corresponding parameters.
• One-click performance optimization
To free the administrators from understanding those sophisticated
performance parameters and tedious performance tuning, the firmware
provides predefined performance profiles for optimizing the storage system
according to different workload. Simply by one click on the GUI, a storage
system optimized for time-critical, transaction-oriented, or high-throughput
applications can be built.