Understanding consistency groups is essential to having a properly functioning disaster recovery process. They speak to the reality that it is not possible to recover everything at the same time. Creating recovery groups makes a recovery from a disaster much less haphazard and less chaotic.
A consistency group is a group of related applications or services that must be recovered together in order to work properly, and this means more than being recovered at the same time. They also typically need to be recovered to the same point in time. Therefore, members of the same consistency group typically have the same RTO and RPO.
Consider, for example, a typical LAMP (Linux, Apache, MySQL & PHP) system, which is a web development platform. In this example, it is a customer order system for a custom print shop where customers upload very large graphics of designs they need printed. Once uploaded, the images need to be easily available to internal people, so the application stores them on an NFS server. Suppose also the environment uses LDAP to manage accounts, and a Linux file server for NFS. For scalability reasons, Linux and Apache are on one server, MySQL is on another server, and the NFS and LDAP services are also each on a separate server. That is a total of four servers, or VMs, that must be recovered together in order for the order system to function.
Consider what would happen if the MySQL database was recovered to one point in time and the NFS server was recovered to another point in time. You would have SQL records of files that don’t exist, or files that exist with no SQL record attached to them. This is referred to as a referential integrity problem. This is why these two applications must be restored to the same point in time.
It’s probably okay, however, to restore the LDAP server to a slightly different point in time, as referential integrity problems will be minimal. But without the LDAP server, administrators would not be able to log into the web and database servers to do their job. Therefore it must be restored at the same time as the other servers in order for the entire system to function.
Within your environment there are likely dozens or hundreds of consistency groups. Many of them probably have overlapping applications, especially when you consider things like directory services. This is why many people restore their directory services first before restoring anything else. Linux environments restore LDAP and Windows environments restore Active Directory and their Primary Domain Controller before doing anything else.
I’ll be talking about consistency groups and the need to decide up front what applications get what type of recovery in our upcoming webinar “How to Future Proof Data Protection for Organizational Resilience“. It’s on Apr 27 2017 at 10 AM PT/1 PM ET. See you there!