Should my primary and replica write to the same shared storage?
09 November 2021 05:21 PM

The purpose of High Availability (HA) in MarkLogic Server (through Local Disk Failover) and Disaster Recovery (DR, through Database Replication) is to make sure you have multiple copies of the same data available so that if one copy is lost, you will have other copies upon which to fall back. 

Consider the example of a cluster with Local Disk Failover (LDF) and Database Replication (DR) configured:

  • You have 5 copies of data available for each data forest:
    • Primary cluster: 2 copies (1 primary forest + 1 LDF forest)
    • Replica cluster: 2 copies (1 primary forest + 1 LDF forest)
    • Backup: 1 copy
  • Specifically, on the primary cluster, forest A (which is a primary data forest - 1st copy) has a local disk failover forest (2nd copy) for high availability. The replica cluster will have its own data forest (3rd copy) for disaster recovery and a local disk failover forest (4th copy) for HA in your DR environment. Backups represent a 5th copy of your forest.

While it is important to make use of features like LDF and DR to achieve high-availability/disaster-recovery through having multiple copies of data, it might be more important to consider how all these copies are stored. In the above example, even though there are 2 copies of data available in the primary cluster (3 copies if you include the backup), if all of these copies are written to the same storage and that single storage environment fails, all of those copies are lost. The best practice in terms of both HA and DR is to ensure that each of your forest's copies lives on its own dedicated storage.

(3 vote(s))
Not helpful

Comments (0)