4. Disk I/O
Disk I/O is one of the most common causes of poor system and application performance. On the latency scale, CPU speeds are measured in gigahertz, memory access takes tens to hundreds of nanoseconds, and network packets make round-trips in microseconds. Disk reads and writes are at the far edge of this time scale, with a typical disk I/O taking several milliseconds. Thus, when we profile application latency, we must measure disk I/O and determine all aspects of a workload’s disk I/O attributes (which files are being read and written, I/O sizes, I/O latency, throughput, and so on) in order to understand application behavior and performance.
DTrace can observe not only details of each disk I/O event but also the inner workings of disk ...
Get DTrace: Dynamic Tracing in Oracle now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
