Putting it all together - a combined topology
An example of a topology that combines the best elements of the other topologies
discussed in this section is shown in the following figure.

This topology combines elements of several different basic topologies:
- Two deployment manager cells
- A deployment manager node to manage each cell (Machine A in cell 1; Machine
J in cell 2)
- Two Load Balancer nodes (Machine B in cell 1; Machine K in cell 2)
- Two HTTP servers for each cell (Machines C and D in cell 1; Machines L
and M in cell 2)
- Four Application Server nodes for each cell (Machines E, F, G, and H in
cell 1; Machines N, O, P, and Q in cell 2)
- The use of cluster members for both vertical and horizontal scaling. In
the example topology, each node hosts three cluster members; in practice,
the number of cluster members is limited by the computing resources of each
node.
- A data tier for each cell.
Typical use
This topology is designed to maximize throughput, availability, and performance.
It incorporates the best practices of the other topologies discussed in this
section:
- Having more than one Load Balancer node, HTTP server, and
application server in each cell eliminates single points of failure.
- Multiple cells provide both hardware and software failure isolation, especially
when upgrades of the application or the Application Server software are rolled
out. You can handle hardware and software upgrades on a cell-by-cell basis
during off-peak hours.
- Horizontal scaling uses both clustering and a load balancer to maximize
availability and eliminate single points of process and hardware failure.
- Application performance is improved by using several techniques:
- Hosting Application Servers on multiple physical Machines to boost the
available processing power.
- Creating multiple smaller cells instead of one large cell. There is less
interprocess communication in a smaller cell, which lets you devote more resources
to processing client requests.
- Using cluster members to vertically scale Application Servers on each
node, which makes more efficient use of machine resources.
- Applications with this topology can use several workload management techniques.
In this example, you can perform workload management through one or both of
the following:
- Using the WebSphere Application Server Network Deployment workload management
(WLM) facility to distribute work among Application Server cluster members.
- Using Load Balancer to distribute client HTTP requests to each Web server.
For example, an application can manage workloads at the Web server
level with Load Balancer and at the Application Server level with WebSphere
Application Server workload management. Using multiple workload management
techniques in an application provides finer control of load balancing.
Regardless
of which workload management techniques are used in the application, Network
Deployment participates in workload management to provide failover support.
In this topology, users notice an interruption only when an entire cell
is lost. If this situation occurs, the active HTTP sessions are lost for half
of the clients. The system can still process HTTP requests, although its performance
is degraded.
The combined topology has several drawbacks:
- Deployment is more complicated. WebSphere Application Server software
and application files must deploy in each cell, which is not the case for
applications that run only in a single cell.
- Multiple cells require more administration effort because each cell is
administered independently. Reduce this problem by using scripting to standardize
and automate common administrative tasks.

Setting up a multinode environment
Searchable topic ID:
cins_together
Last updated: Jun 21, 2007 8:07:48 PM CDT
WebSphere Business Integration Server Foundation, Version 5.0.2
http://publib.boulder.ibm.com/infocenter/wasinfo/index.jsp?topic=/com.ibm.wasee.doc/info/ee/ae/ctop_together.html