Q: What are my disaster recovery options for a System Center Configuration Manager 2012 site if I lose all servers in the site?
A: I had a customer with a primary data center who was looking at options for disaster recovery for a primary site that had all site systems and most importantly the database in the primary data center. The question was how to make the site highly available, or more specifically, disaster-recovery enabled.
A System Center Configuration Manager primary site consists of the site database, which is hosted on SQL Server (and can be a SQL Server cluster) and roles such as Management Points and Distribution Points. There is no support for using SQL Server AlwaysOn to replicate the Configuration Manager database to another SQL Server installation (e.g., at another location—i.e., the disaster recovery location). Within a site, you achieve high availability by having multiple instances of roles, such as multiple Management Points and multiple Distribution Points, and ensure the site database is on a SQL Server cluster. However, the same approach doesn't work for disaster recovery across different locations, because of the SQL Server database.
There are therefore three main options (in order of preference):
- Perform a site recovery at the disaster recovery location, which consists of standing up a new Configuration Manager server (named the same as the server that failed). Next, you reinstall Configuration Manager but select the Recover a site option and select the location of your backup for the site and the location of the backup of the SQL Server database. (This assumes you have a backup task that's regularly backing up both the Configuration Manager configuration and the SQL Server database. One option is to use the built-in Site Maintenance - Backup Site Server task that's disabled by default.)
- An unsupported but commonly used option is to virtualize the various roles, including SQL Server, and then use virtualization HA/mobility to move the virtual machines containing Configuration Manager site systems in the event of a disaster. This could include leveraging Hyper-V Replica.
- Have another primary site at the disaster recovery location, and in the event of a disaster move all the clients to the disaster recovery primary site. However, this causes a lot of extra work on the clients and traffic, including when you have to move them back.
