Solid State Disk – How it Works?
A single solid state disk can handle about 250 times more small block random I/O's per
second than a hard disk drive.
That is, 250 disk drives would be needed to perform the same amount of work as one SSD.
Virtually any I/O bound application would benefit from being placed on an SSD.
Many Solid Access customers place entire applications on a USSD 200, but that is
not always feasible. A significant application acceleration can be achieved by
off-loading only I/O-demanding files (“hot files,” typically less than 5% of the
content) onto an SSD for processing at RAM speed and using mechanical disks
(RAID or JBOD) to process the remaining “cold files.” This instantly improves
the efficiency of the application servers by recovering CPU cycles formerly
lost in I/O wait loops.
Database Acceleration
The approach above is used for high end database acceleration in many industries.
Any database component that creates I/O bottlenecks, such as temp tables, log files
and indexes, should be placed on SSD. These files often represent very small part of
database capacity (5%) but account for more then 50% of I/O bottlenecks. Solid state
disks have a strong following within the Oracle and SQL Server communities.
“Specifically these applications have shown tremendous performance increase
by the addition of Solid State Disk: DB2's Log files, Indices, Temp table. Informix:
RootDBS table, PHYSDBS, LOGDBS, and TMPDBS1. Oracle: Redo logs, Rollback Segments, and
Temporary tables. Progress: AI, BI, and Temporary tables. Sybase: Transaction Logs,
Tempdb, heavily hit tables. SQL: Heavily hit tables, Transaction logs, and Temporary
tables. SAP: PSAPBTAB tables, PSAPCLU tables, PSAPSTAB tables, M tables.”
Tier 0 - Are you ready? Storage Switzerland, 5/15/07
Storage Area Network Acceleration
The arrival of the tiered storage concept used in SAN switched fabric environments enables efficient utilization of available resources. Ultra fast SSD is an ideal top performing resource
(Tier 0) in SAN for random I/O hungry application segments.
Evolving virtualization simplifies a proper tier assignment based on an application I/O segment profile.
It enables storage resources to be paired dynamically with an application segment based on its I/O temperature. Policy-based data migration ensures that the right data is in the
right place at the right time.
|
