DB2 II: Performance Monitoring, Tuning and Capacity Planning Guide

26 DB2 II: Performance Monitoring, Tuning and Capacity Planning Guide

1.5 DB2 Information Integrator topology considerations

Organizations choose Information Technology (IT) platforms and topologies

based on their application characteristics, scalability, and availability

requirements. For example, for an online stock trading Web application, an

organization might choose the AIX platform for the WebSphere Application

Server (WAS) and implement both vertical and horizontal WAS clones to achieve

the desired scalability and availability requirements of the application.

The key decisions to be made for a DB2 II implementation are:

򐂰 What platform to choose: UNIX, Windows, or Linux?

򐂰 Dedicated federated server or variations of collocation with data sources?

򐂰 What are the capacity requirements of DB2 II?

In this section, we briefly describe the basic topologies and discuss the key

criteria involved in choosing between them.

Figure 1-6 on page 27 shows the two basic topologies involved, as follows:

򐂰 The dedicated federated server has no data sources in the same machine as

the federated server whatsoever.

򐂰 There are two variations of the collocated federated server, as shown in

Figure 1-6 on page 27:

– The DB2 data source is enabled to be the federated server. There may or

may not be other DB2 and non-DB2 relational data sources on the same

machine. The machine may also house non-relational data that may be

referenced in one or more federated queries.

– The federated server is in a separate DB2 instance/database on a

machine that has DB2 or non-DB2 relational data sources sharing the

machine. The machine may also house non-relational data that may be

referenced in one or more federated queries.

In general, the DB2 data source enabled to be the federated server option

performs better than the option where the federated server has its own DB2

instance/database, especially when the database is partitioned (DPF).

However, the federated server in its own DB2 instance/database provides

better fault tolerance and isolation.

Chapter 1. DB2 Information Integrator architecture overview 27

Figure 1-6 DB2 II topologies

The key criteria in choosing a particular DB2 II topology are scalability,

availability, total cost of ownership, performance and access considerations.

Attention: The choice of a particular platform (UNIX, Windows or Linux) is

largely driven by scalability, availability, and total cost of ownership

considerations, and will not be discussed here.

Estimating the CPU and memory capacity requirements of a federated server

for a new application is also beyond the scope of this book. However, for

organizations that currently have DB2 II installed, Chapter 5, “Capacity

planning in an existing DB2 II environment” on page 377, documents a

procedure for developing custom query utilization profiles that may be used to

estimate the capacity requirements of new applications on the same or

different federated server.

UNIX, Windows, or Linux

Local

Non-relaional

data

COLLOCATED

Federated Server

sharing same DB2 database

as the DB2 data source

DB2 Information Integrator

Federated

Server

Non-relational

Data Sources

Relational

Data Sources

DEDICATED

Federated Server

Federated

Server

+

DB2 data

source

UNIX, Windows, or Linux

DB2 Information Integrator

Non-relational

Data Sources

Relational

Data Sources

UNIX, Windows, or Linux

Local

Non-relaional

data

COLLOCATED

Federated Server

in a different DB2 instance/database

than the DB2 (or non-DB2) data sources

DB2 Information Integrator

Non-relational

Data Sources

Relational

Data Sources

Federated

Server

DB2 or

non-DB2

data sources

DB2 or

non-DB2

data sources