Posted on 2011-03-21
This question isn't so much to determine a correct answer, but more gauging opinions as to the right course of action.
The 2007 Exchange server at this organisation has multiple ways of recovering data. The mailbox server role is clustered across 3 servers. Each server points to a single SAN containing the mailboxes and logs for exchange, and that SAN is replicated to another SAN that is offsite in case of a DR scenario. Mail is also archived in the cloud before reaching this Exchange server and all mail sent internally is processed through journaling and sent to the cloud providing full availability should this Exchange server go down.
Although the solution above provides high availability and the ability to restore the data should one of the SANS go down, transaction logs are not committed to the database and thereby not flushed. To achieve this, monthly archives are done by backing up the databases to a disc locally.
What are people’s opinions on using circular logging in this situation instead of backing up? For example, is the culmination of log files still necessary, or is it ok to overwrite the existing log files in a live environment? Are there performance issues people are aware of that come to light after enabling circular logging for long periods of time?
Any help and opinions would be most helpful.