Terminology
The following terms are used in this guide.
Load Balancing
Load balancing is a technology for forwarding client requests to more than one server host machine. Having multiple servers available to handle client requests typically improves an application’s overall throughput capacity. For more information, see:
Load Balancing in the Microsoft MSDN library
Service
A service is an application that executes on a particular machine that can host servers. Services provide methods to start and stop applications, install and uninstall them, and otherwise manage their execution lifetimes. You can have Windows services and web services. Two examples of Windchill Risk and Reliability services follow:
The Object Broker Service is a Windows service. It establishes communication between Windchill Risk and Reliability applications and databases. It also provides login and permission checking for applications.
Server
A server is an instance of an interface that is hosted within a service. The server is responsible for providing both an interface that can be invoked and an implementation of that interface. The following is an excerpt from a Windows Communication Foundation (WCF) configuration file for Windchill Risk and Reliability. It provides an example of an interface, or contract in WCF terms, and an implementation.
* 
Many internal component names for Windchill Risk and Reliability still include Relex which was the product suite name before PTC rebranding.
Singular Support Servers
While most Windchill Risk and Reliability servers can be load balanced, RelexBusinessObjects.SingularLRTServer and RelexBusinessObjects.CacheManager Server are singular support servers, which means that they can exist in only one location. Hosted in the Long Running Task Service, these two servers cannot be hosted on a machine that is load balanced. This information is also provided in the server-side configuration file that gets installed: RiskAndReliability.Service.LongRunningTasks.exe.
Benchmarks
Because Windchill Risk and Reliability uses a multi-tier distributed architecture, more than one service is running on the application server. Making load balancing available in multiple places should enhance overall system performance. However, if load balancing is not required, it can be detrimental to system performance. Consequently, you should always run benchmarks before and after load balancing to ensure performance gains are realized.