Performance Tuning for Content Manager

328 Performance Tuning for Content Manager

increase of transaction throughput and a reduction of transaction length

(measured in time units). Thus, locks are not held as long as before.

In this example, we just monitored the Library Server database. Without going

into further details, we note that Performance Expert enables you to open

windows that provide graphical views to simultaneously monitor the Library

Server database and the Resource Manager database.

Recommendations for using PE to monitor CM databases

DB2 Performance Expert Server is a sophisticated monitoring and analysis

application that is built on DB2. This means that it adds a certain load to the

server machine.

When using it, we recommend avoiding resource contention and keeping

monitoring completely separated from your Content Manager server workload.

You should install the Performance Expert server on a different machine than the

Library Server and the Resource Manager.

11.4.3 DB2 analysis tools

In 11.4.2, “DB2 monitoring tools” on page 309, we give an overview of the tools

that you can use to monitor a database environment in order to identify a problem

if there is any. You might not be able to retrieve sufficient information from the

monitoring tool to solve some problems. In some cases, you might need to

perform further analysis at the operating system level (for example, if it comes to

the impact of low buffer pool hit ratio on disk I/O) and in other cases, you may

need to use DB2 analysis tools. This is typically the case when you figure out that

execution of a specific SQL statement is very slow and consumes unreasonably

high CPU or I/O resources in your database system. In the DB2 Explain section,

we describe recommended steps toward identifying the cause of the problem

and helping you find a way to solve it.

We discuss two DB2 analysis tools here:

 DB2 Explain

 DB2 Design Advisor

DB2 Explain

As described in 2.1.3, “Performance-related concepts” on page 39, DB2 involves

the SQL query optimizer to evaluate and compare the cost of access plans to

finally choose the access plan that appears to be most efficient. Nevertheless,

even when using this best access plan, the query execution might turn out to be

very expensive. This is typically observed by the user as long response time or

by the database administrator as a high load on disks and network. Using the

monitoring tools described earlier in this chapter, you will be able to identify this