One of the major changes between Lync 2010 and Lync 2013 is the integration of Lync with Windows Fabric. Windows Fabric is described as ‘a Microsoft technology used for creating highly reliable, distributable an scalable applications’. Windows Fabric is designed to enable replication of application related data across multiple application servers. These application servers work in parallel to provide fault tolerance and scalability. Windows Fabric ensures the information needed for these applications is delivered reliably to each of the necessary servers.
Lync uses Windows Fabric to move information previously held in a back end database (and hence a bottleneck) to the front end servers. This eliminates the need for each Lync Front End server to use network resources to obtain application information – it’s available locally.
In the case of Lync, this change enables better scaling as well as providing resilience (a front end server can continue to work in the case of errors elsewhere in your network). While pretty much all of the complex stuff, with respect to Windows Fabric, is done auto-magically for you, it is important to understand what it’s doing, why and some of the issues arising.
One of key issues that arises from the integration of Windows Fabric and Lync is the recommendation of always having at least three ES servers in an ES pool. See http://technet.microsoft.com/en-us/library/gg412996.aspx for the details of this limitation.
For more information on the integration of Lync 2013 and Windows Fabric, see Richard Brynsteson’s post on the Mastering Lync blog.