Oracle Disk Monitoring Oracle Tips by Burleson Consulting

Oracle Disk Monitoring

Everyone has better things to do than sit around and run disk performance scripts.  The process can be automated using scripts and the cron tab in UNIX and Linux. A simple web search using any search tool of choice, one can find any number of shell, perl and other scripts to monitor using iostat , vmstat and sar .

Rather than recreating the wheel, the web can be used to find examples of scripts. Mike Ault’s book, Oracle Disk I/O Tuning (2004, Rampant TechPress), also provides a plethora of scripts and examples.

Tuning professionals have noted that the primary bottleneck is disk I/O for the vast majority of non-scientific systems. Back in the days before RAID and giant db_cache_size, the DBA had to manually load balance the disk I/O subsystem to relieve contention on the disks and the disk controllers.

However, is it really important to find a hot disk? Unless the system is running in single disk sets (jbod technology), specific knowledge about a hot disk is not quite as important. However, the DBA should still be interested in disk performance and tracking data file and other I/O involved in databases. The database administrator and system manager must still monitor I/O, either through Oracle, through the OS, or via performance monitoring tools provided by the disk array or storage system software vendors such as Veritas.

Many DBAs would like to believe that this disk technology has changed. Unfortunately, the only major changes to disk technology since the 1970s are these hardware and software changes:

Large data buffers: Today the DBA can cache large portions of the data blocks in the db_cache_size reducing disk I/O.

Disks with on-board cache: Many of the newer disks have an on-board RAM cache to hold the most frequently referenced data blocks.

RAID: The randomizing of data blocks with RAID 1+0 and RAID 5 for low-write systems has removed the need for disk load balancing by scrambling the data blocks across many disk spindles. In Oracle10g, the Automatic Storage Management (ASM) feature requires Stripe and Mirror Everywhere (SAME) which is essentially RAID 1+0 and RAID-10.

Other than these advances, basic disk technology has not changed since the 1970s. The Oracle DBA must remember that disk I/O remains an important issue and understand the internals of disk management to maximize the performance of their I/O-bound systems.

A vendor-specific disk-level I/O monitor

When Oracle monitoring is used, at least in 9i and lower versions, DBAs were somewhat limited in what monitoring could be performed. Total I/O since startup and I/O/second since startup can be captured.  This information could be acquired on a per datafile or tempfile basis. The v$filestat and v$tempstat dynamic performance views provide the I/O statistics from Oracle’s point of view.

---

This is an excerpt from my latest book "Oracle Tuning: The Definitive Reference".  You can buy it direct from the publisher for 30%-off and get instant access to the code depot of Oracle tuning scripts: