I have been asked this question quite a few time now.
As much as I believe (according to a few SQL administrations) it is best practice to match the server collation with the database collation for performance reasons for Lync you really don’t want to this.
The server collation should be Latin1_General_CI_AS
If you interested all the Lync Server 2013 databases use Latin1_General_BIN collation, this is a case sensitivity collation and if you happen to match the collation then you will probably run into issues with mirroring the archiving and monitoring databases as they their database naming is mixed case.
