Messaging is a business critical application for any enterprise. To increase the productivity, instant or real-time communication is very much required. Lync has become a more suitable option for business productivity which serves all real-time workloads such as IM/Presence, Audio/Video & Web conferencing, Group/Persistent Chat, Application & Desktop sharing and Enterprise Voice.
While enterprises are moving towards an application for real-time communication, services availability is concern for IT decision makers. To avoid complexities and provide more availability Microsoft came up with brick model architecture for Lync 2013. It allows scaling out your Lync FE servers from 1 to 12 and each server can support up to maximum 6666 concurrent users. Lync Server 2013 is loosely coupled with Lync Back End SQL server and use lazy writes feature to write data in SQL Back End server.
Lync Server 2013 divides users in a group and each group is dynamically assigned to three Front End Server, primary secondary and tertiary. Once user logs in, primary server serves the services and replicate the data across secondary and tertiary using synchronous replication as well as update the SQL backend server using asynchronous replication known as lazy writes. In case of primary server failure secondary server serves the services.
In Lync server 2013, Front End server keeps rtc, rtcdyn and xds database in local sql express store. To know more about Lync 2013 databases please go through Lync Server 2013 databases.
Now, we should talk about how brick model works. Basically, brick model is based on windows fabric and it consider nodes majority for availability. That’s why you need minimum three Front End servers to achieve high availability. The majority is considered when n/2 or n+1/2 Front End servers are alive.
For maintenance, you should restart/down one server at a time. If all Front End servers go down due to the power outage or any other reasons, please make sure you are starting all Front End servers at the same time.