AFM DR deployment considerations and best practices

This document describes the characteristics of AFM DR including advantages and limitations, its deployment method, and best practices.

AFM DR is a file-level asynchronous disaster recovery solution. Replication of data to a secondary cluster is asynchronous, which means that the data created on the primary cluster is copied to the secondary cluster with delay. The amount of delay is determined by the network bandwidth between the sites and some IBM Spectrum Scale™ tuning parameters.

| AFM DR copies only the data blocks that have changed on the primary cluster to the secondary cluster, which makes it a very efficient replication mechanism.

| Since data replication is asynchronous using AFM DR to recover data after a primary cluster failure might cause some data loss. The amount of data lost depends on the amount of data that is in the primary queue when a failure occurs. Therefore, AFM DR is not the right disaster recovery solution if your application cannot tolerate possible data loss. If this loss is not acceptable,customers should implement a synchronous disaster recovery mechanism.