IBM Tivoli Monitoring: Implementation and Performance Optimization for Large Scale Environments

Chapter 2. Planning considerations - large scale deployment of Tivoli Monitoring V6.1 components 31

2.3.1 Where to run the hub Tivoli Enterprise Monitoring Server

Consider the following seven factors when deciding where to run the hub

monitoring server. Not all factors will have the same significance in every

environment; some will be decisive, while others may not matter at all to you.

Portal user authentication

Even though they logon to the portal server, users attempting to use the Tivoli

Enterprise Portal application must be authenticated at the hub monitoring server.

The hub monitoring server calls the local security authority to confirm that the

userid and password are valid (if User Validation is enabled), but the

authorization table (what the user can do) is still kept in the portal server

database, KFWUSER table.

Where the userid is defined depends on the platform where the hub monitoring

server is running, as explained here:

򐂰 On z/OS, this will be System Authorization Facility (Resource Access Control

Facility (RACF®), Access Control Facility2 (ACF2), or Top Secret Security).

򐂰 On Windows, the id could be local to the hub monitoring server machine or

defined in the Domain to which the hub monitoring server machine belongs

(or a trusted domain).

The id must be granted the “Logon locally” right to the hub monitoring server

machine. On UNIX or Linux, this will be local security until support for LDAP is

implemented.

Except for XE for Messaging, the userid is only used to logon to the portal server

and will not be used again for the life of the session. With XE for Messaging, the

userid of the portal user can be used to control access to view message contents

or issue commands (Take Action) against the Queue Manager.

Note: Part of this section is based on a whitepaper written by Richard Roy

from IBM Canada. The whitepaper entitled "ITM Components Relocation" is

published at

http://catalog.lotus.com/wps/portal/topal/details?catalog.label=1TW1

0TM68.

Note: IBM Tivoli Monitoring V6.2 will provide LDAP support.

32 IBM Tivoli Monitoring: Implementation and Performance Optimization for Large Scale Environments

Availability

Availability takes two forms: availability of the platform itself, and availability of the

hub monitoring server as an application running on it.

z/OS is generally viewed as highly available, but with higher exploitation of

Parallel Sysplex®, availability of any single LPAR is less guaranteed than that of

the sysplex. At the limit, any LPAR is expendable because its workload can freely

move to any other LPAR. The risk of viruses and attacks and the frequent

distribution of patches can make Windows platforms less attractive.

ASCII/EBCDIC data conversion

Data conversion refers to the need to convert data moving between ASCII and

EBCDIC machines. If the hub monitoring server and a component sending data

to it are on different platforms, the receiving end of any transmission will need to

convert the data stream to the local character format. Depending on the amount

of traffic, this could become significant.

Note that the entire presentation layer of IBM Tivoli Monitoring runs on ASCII

platforms, so data will eventually have to be converted to ASCII if it is to be

viewed by end users. Data conversion is often overlooked in deciding where to

place the IBM Tivoli Monitoring hub monitoring server.

Event Forwarding

Event Forwarding is intended to simplify the integration between IBM Tivoli

Monitoring and other Tivoli products, specifically the Event Management

components of IBM Tivoli Enterprise Console, Netcool/Omnibus and IBM Tivoli

Business Systems Manager. It is part of a larger integration called Event

Synchronization. This feature is currently not available when running the hub

monitoring server on z/OS.

Although it is possible to forward events to IBM Tivoli Enterprise Console,

Netcool/Omnibus and IBM Tivoli Business Systems Manager from a z/OS hub

monitoring server, the process is more complicated and may require the use of

Workflow Automation (Policies) and additional products such as z/NetView or the

Universal Agent.

Note: For z/OS hub high availability considerations, you can refer to the

"High-Availability HUB TEMS on z/OS" whiteaper at

http://catalog.lotus.com/wps/poral/topal/details?cataog.label=1TW10T

M61.