Best Practices

There are several items recommended in this chapter that should be taken into account when planning a non-voice deployment of Lync Server 2010. By following these recommendations, you can be better prepared for the deployment and can avoid the common pitfalls associated with planning a topology and deployment of a complex technology. The following is a summary of recommended best practices from this chapter.

• Start the deployment planning with a comprehensive design document. This ensures that decisions have been made, and it gives an excellent opportunity to shop the design around to other groups to get buy-in and to ensure that other groups know how they’ll be affected by the upcoming deployment.

• Treat your deployment like a formal project. Produce a project plan that includes anticipated tasks and anticipated durations, and calls out the required resources. This makes it easier to get support from management to dedicate the appropriate resources and time to the deployment.

• Make sure that any constraints are understood prior to the completion of the design. Things such as regulator compliances have a major impact on decisions made in the design.

• Make sure the deployment can handle the anticipated load. Use the Microsoft sizing guidelines to ensure that the design can support the load you expect to place on it.

• Start the deployment with a pilot. This gives an excellent opportunity to validate the impact of users on the system and gives administrators an opportunity to get familiar with the infrastructure while supporting only a limited number of users.

• Involve the networking group when designing a non-voice deployment. The new features offered by Lync Server 2010 will greatly impact a WAN and the networking group will have bandwidth and latency information that might affect the design.

• Make decisions about how external users will be supported early in the process. Designs for internal only versus external support will be significantly different. These decisions affect several other decisions, so the earlier this can be decided, the less impact it has on the overall effort.

• If possible, use public certificates that use subject alternate names. This greatly reduces the impact on end users and on external users because the certificates will already be trusted by the operating system.

• Use modern hardware. Current-generation processors can provide a much larger capacity than processors from only one generation ago. The Microsoft sizing guides are based on current generation processors. Deviating from that practice renders the sizing guides inaccurate and puts the project at risk of being underpowered. This results in a poor user experience when loads increase.

• Train the end users on how to use the new system. Ensure that the users understand the limitations of various clients because they do not all provide the same features. Consider an online FAQ (Frequently Asked Questions) for users to refer to.

• Define an acceptable use policy for the Lync Server 2010 system. This reduces the exposure of the end users because they will know what behaviors can potentially put them at risk. It also enables administrators to more easily block risky behaviors because there will be a written policy to back up the configuration.

• Make sure the archive is designed with enough storage to hold the anticipated volume of data.

• Regularly test the recovery of data from the archive. This helps ensure that the data will be available and readable when the time comes that it’s needed for something important. Consider a monthly test.

• Plan carefully when using virtualization and understand what other services are provided by the virtualization farm. Most virtualizations farms are oversubscribed because virtualization is a popular form of consolidation for servers. Be aware that not having dedicated resources reduces the potential capacity of the systems.

• Whenever possible, use a system of role based access control. Roles should be well defined and administrators should be given only the minimum of rights needed to perform their jobs. If their jobs change, their role-based group memberships should change. Don’t forget that this goes both ways: Add them to new groups when they need additional rights, but don’t forget to remove them from groups if there are tasks they longer perform.

• Don’t forget to monitor the Lync Server 2010 environment with an application, such as SCOM or Nagios or SiteScope, to ensure that the systems are available. Leverage the monitoring software to trend the loads on the system to be able to predict when extra capacity will need to be added to maintain an acceptable load on each server.

By following these best practices, administrators can maximize their chances for a successful Lync Server 2010 deployment and can keep their end users happy and productive.