
396 DB2 UDB ESE V8 Performance Guide for High Performance OLTP and BI
6.1 Introduction
Most IT environments today are complex infrastructures involving heterogeneous
network, hardware, and software components organized in multi-tier
configurations. Business applications share this complex IT infrastructure, which
is managed by IT professionals skilled in their particular domain of expertise, for
example, network administrators, operating system administrators, Web
application server administrators, and database administrators.
As mentioned earlier, users may experience performance problems for reasons
such as network connectivity and bandwidth constraints, system CPU, I/O and
memory constraints, software configuration limitations and constraints,
inadequate systems administration skills, poor application design, and faulty
assumptions about the workload. In this chapter, we address the following
related topics:
1.4, “Problem determination methodology” on page 7 discusses a general
problem determination methodology, and recommends a hypotheses validation
hierarchy that should typically be followed during problem diagnosis of DB2
applications in general.
For the DB2 application environment shown in Figure 6-1 on page 397, the
diagnosis process should sequentially eliminate the cause of the problem as
follows:
1. Network-related - between the client and the Web application server
2. Web application server-related - both system (CPU, I/O, memory) and various
configuration settings
3. Network-related - between the Web application server and the database
server
4. Database server-related - system (CPU, I/O, memory), configuration settings,
and routine DBA maintenance activities such as collecting statistics or
reorganizing tables
5. Application design-related - tables and SQL
Important: Following this sequence is strongly recommended in order to
ensure that the DBA does not expend needless effort on troubleshooting DB2,
when the root cause of the performance problem experienced by the user
potentially exists elsewhere. For example, network bandwidth constraints or
resource contention in the Web application server can manifest as erratic or
poor response times for a user of a DB2 application even when the DB2
system and application is perfectly tuned.