Enterprise Library Test Guide - Willy .Net
Enterprise Library Test Guide - Willy .Net
Enterprise Library Test Guide - Willy .Net
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
216<br />
<strong>Enterprise</strong> <strong>Library</strong> <strong>Test</strong> <strong>Guide</strong><br />
You should monitor I/O performance on the computer that hosts the application<br />
block, the computer that hosts SQL Server, and on any other computer that hosts a<br />
resource that the application block uses. Table 25 lists the performance counters you<br />
can use to analyze disk I/O activity.<br />
Table 25: Disk I/O Performance Counters<br />
Performance monitor counter<br />
Description and recommendations<br />
% Disk Time This is the percentage of elapsed time that<br />
the selected disk drive is busy servicing I/O<br />
requests. This should be approximately 5<br />
percent.<br />
% Idle Time This is the percentage of time during the<br />
sample interval that the disk was idle. This<br />
number should be approximately 95 percent on<br />
both the Web server and the SQL Server.<br />
Disk Reads/Sec and Disk Writes/Sec<br />
Together, these counters represent the number<br />
of I/O operations issued against a particular<br />
disk. Generally, there is a practical limit of<br />
100 to140 operations per second per spindle.<br />
Consult with your hardware vendor for a more<br />
accurate estimation.<br />
Avg. Disk sec/Read and Avg. Disk sec/Write Together, these counters measure disk latency.<br />
Lower values are better than higher values, but<br />
this value can vary and is dependent on the<br />
size of the I/O operations and the workload<br />
characteristics. Numbers also vary across<br />
different storage configurations. For example,<br />
the storage area network (SAN) cache size and<br />
how often the cache is used can greatly impact<br />
this metric.<br />
On well-tuned online transaction processing<br />
(OLTP) systems that are deployed on high performance<br />
SANs, the ideal values vary between<br />
less than 2 ms for the log files and 4 ms to 10<br />
ms for data. Decision support system (DSS)<br />
workloads may have higher latencies of 30 ms<br />
or more. For Internet Information Services (IIS),<br />
the number should be between 8 ms and 10<br />
ms.<br />
Persistent values of greater than 100 ms can<br />
indicate I/O problems. However, this value is<br />
dependent on workload characteristics and<br />
the system hardware. When considering this<br />
measurement, keep in mind the normal values<br />
for your system.