How Lync Server Leverages Multiple SQL Instances for High Availability

SKYPE for Business (Lync) Server uses multiple database instances and types: Active Directory stores basic user information such as SIP URI and phone numbers, while policy information is kept in the CMS database (SQL). In addition, the RTC and Rtcdyn databases are also used.

Although a single SQL instance could theoretically host the backend, Lync actually deploys several SQL Express instances. Each Lync server—including front‑end servers—runs its own SQL Express instance that holds copies of backend databases (CMS, RTC, Rtcdyn).

Replication is performed via push replication: a CMS master pushes data to each Lync server’s replica replicator agent, which receives the data and reports back to the master. This push model works like a broadcast, updating all servers quickly, unlike pull replication that requires each server to fetch data at intervals.

This design improves availability: if the backend SQL fails, other Lync servers still have local copies of the configuration, so most Lync functions continue to work. Consequently, deploying the backend as a mirrored or clustered SQL instance is optional and should be weighed against the added maintenance complexity.

The typical deployment sequence is: prepare Active Directory, set up the CMS SQL database, use the topology generator to create configuration data, and finally install or update Lync based on that configuration. The standard edition uses SQL Express for the CMS store, while the enterprise edition employs a dedicated backend SQL server.

In the standard edition, most Lync roles can be installed on a single server, which is advantageous for small‑to‑medium environments. The enterprise edition separates roles across multiple servers, making the architecture more complex.

Key SQL instances include:

RTC instance – the backend database.

RTCLOCAL instance – present on each Lync server.

LYNCLOCAL instance – also present on each server.