||Oracle Tips by Burleson
chasm between bandwidth and performance
In addition to the growing gap between storage
performance and processor performance is a similar chasm between
bandwidth and storage performance. Bandwidth is becoming available at
rates that exceed even the improvement in processor performance.
Already today, there is a glut of bandwidth that is now available to
enterprises for a much lower cost than was ever possible. With almost
unlimited bandwidth, the demands on the remainder of the data center
only intensify. From the firewall to storage, every aspect of system
performance will need to be analyzed for bottlenecks.
Storage densities are increasing, and data access
problems are only getting worse. Going forward, storage managers will
be required to balance the need for capacity with the need for
performance. There are no easy solutions to the problem if the options
only include hard disk based JBOD or RAID systems. It is the
mechanical aspect of hard drives that ensures they will always be much
slower than the server or incapable of filling massive enterprise
The next section explores how these advances in
hardware are shifting Oracle’s processing bottlenecks.
The shift in Oracle
As noted in Chapter 1, the dramatic decrease in
the cost of RAM is radically changing the Oracle database
architecture. In 2005, 100 gigabytes of SSD (i.e. Texas Memory
Systems) can be obtained for about $150,000.
Historically, RAM has been a scarce and expensive
resource, and the DBA was challenged to find the data that was
referenced most frequently to cache on precious RAM media.
This new age of cheap solid state disks has meant
a dramatic change in Oracle database architecture, as the old
fashioned model of disk-based data management is being abandoned in
favor of a cache-based I/O approach.
For years, storage architects have observed the
growing divide between processor performance and storage access
times. Remember, when the CPU waits on storage, the users are waiting
According to a whitepaper by James Morle, SSD is
great for high-bandwidth I/O components such as Oracle redo logs.
The Oracle redo logs archive the row before images and they are used
for data recovery in the event of a disk crash. Morle notes:
“This is where SSD
can help out. By deploying a single SSD, all redo logs can be
located away from the RAID 1+0 array, whilst providing low latency
writes and high bandwidth reads (for archiving).”
The next section looks at the inherent bandwidth
limitations of platter-based disk storage so that a better
understanding can get gained of the benefits of SSD for Oracle data
The above book excerpt is from:
Oracle RAC & Tuning with
Solid State Disk
Expert Secrets for High
Performance Clustered Grid Computing
Donald K. Burleson & Mike Ault