Example large deployment architecture

Use this example to familiarize yourself with the architecture of a large Network Manager deployment.

Components

This example deployment consists of:

  • One ObjectServer and one Network Manager installation in London. The London domain sends events and topology to San Francisco.
  • One ObjectServer and one Network Manager installation in New York. The New York domain also sends events and topology to San Francisco.
  • One ObjectServer and one Dashboard Application Services Hub installation in San Francisco. The ObjectServer in San Francisco consolidates the events from London and New York. The Dashboard Application Services Hub server in San Francisco can access topology from both London and New York, but does not consolidate the topologies. Clients anywhere in the world can connect to the Dashboard Application Services Hub server, and view topology from London and New York.
The following figure shows the architecture for this deployment.
Note: For a large deployment of this sort, network latency across the WAN should be taken into account. This is especially important if if the poller is expected to store a lot of historical data. .
Figure 1. Large deployment architecture
Large deployment architecture

Allocation of host workstations

The following figure shows an example allocation of servers for this deployment.

Figure 2. Large deployment host machine allocation
Large deployment host machine allocation

Steps to install a large deployment

The following steps provide an overview of the tasks required for this deployment, and help plan for a similar deployment. The actual steps are detailed in Installing Network Manager and related products and Configuring Network Manager.
Note: If you are installing distributed Network Manager core component and Web application servers with different time zones, then you should set the same time zone on all servers which includes the database server as well as the core and GUI servers. This ensures that Network Manager is able to perform accurate timestamp comparisons from processes on different servers. You should also advise end users, such as network operators, that the system might display times that are different to the time in their location.

To install this deployment, perform the following steps:

  1. Install the topology database on San Francisco host machine 3, and create the necessary database tables.
    Note: The topology database must be installed and started before you start the Network Manager core components so that discovery data can be saved.
  2. Install the following ObjectServers and related components:
    • Install the ObjectServer on San Francisco host machine 2.
    • Install the ObjectServer and the uni-directional gateway on London host machine 2.
    • Install the ObjectServer and the uni-directional gateway on New York host machine 2.
  3. Configure and run the ObjectServers.
    Note: The ObjectServers must be running before the Network Manager core components are started.
  4. Install the Network Manager core components on London host machine 1.
    Note: The Network Manager core components must be installed before the Web applications.
  5. Install the Network Manager core components on New York host machine 1.
  6. Install the Network Manager GUI framework on host machine 3. The GUI framework includes the following software:
    • WebSphere Application Server
    • Dashboard Application Services Hub
    • Network Manager GUI components
    • Tivoli Netcool/OMNIbus Web GUI
    • Cognos Analytics
    Tip: If you install the Dashboard Application Services Hub on a machine with no other products, performance is likely to be better than if you install it on a machine with other products.
    Note: The Network Manager core components must be installed before the Network Manager GUI framework.