By providing fault tolerance in your Microsoft Provisioning System (MPS) environment, you increase system availability, which ensures reliable and streamlined customer service. By distributing network traffic among redundant MPS components, load balancing allows your system to operate efficiently without the loss of provisioning functionality.

The advantages of this type of Service Provisioning configuration are that you can:

• Provide a comprehensive provisioning solution for customers that includes Hosted Exchange, Exchange Mobility, DNS, Office Communication Services, and Windows SharePoint services, while making use of preconfigured customer service plans.
  
• Create a reliable provisioning environment with high availability and that is resistant to faults and failures.
  
• Create a secure provisioning environment that enables you to safely delegate administrative tasks to customers or other administrators in the service provider environment.
  
• Employ automation to streamline multistep provisioning processes within your data center.
  
• Use the MPS Deployment Tool, which is a single installation application, to automate the deployment and configuration of all MPS components while minimizing the occurrence of errors.
  

When planning for fault tolerance and load balancing, you need to consider the following issues:

• The number of provisioning front end servers that will be submitting XML requests into MPS.
  
• The number of Provisioning Engines you will make available for load-balancing XML requests generated by provisioning front end servers.
  
• The number of Transaction log databases you will have for load balancing Provisioning Engine traffic and mitigating single point of failure (also includes the number of Audit and Recovery Services databases).
  
• The total number of servers required to support these configurations
  

The concepts you will need to understand when planning for load balancing and fault tolerance in your MPS deployment are included in the sections that follow.

As part of your planning process, you should strongly consider creating a test configuration that enables you to perform a trial deployment outside your production-provisioning environment. To facilitate testing, you might need to determine if you have available lab space or a network location where you can validate the deployment process and initial MPS configuration.

Because of the complexity of the fault tolerant MPS configuration, it is advised that you implement a trial MPS deployment to become familiar with using the MPS Deployment Tool to design a deployment configuration that distributes redundant MPS components across numerous servers.

See Create an MPS Test Configuration for further information.

The basis for a fault tolerant service provisioning deployment is a clustered SQL Server to reliably host the provisioning databases and the Provisioning Audit and Recovery Service. Using a database server cluster has the advantage of allowing you to add up to 16 provisioning engine servers to the environment without requiring changes to the provisioning database configuration. You need multiple provisioning front end servers to be able to make the best use of having multiple provisioning engines. The minimum fault tolerant service provisioning environment contains Minimal Fault Tolerant Server Roles six servers, including clustered provisioning database servers.