Sometimes it is necessary to change or update a data replication engine. A zero downtime migration will impose no downtime on either applications or users. The migration can take place during normal working hours and occur over an extended time, while the backup database is always available and fully synchronized with the production database. Therefore, application availability is ensured during the migration process. This migration technique is similar to HPE Shadowbase Zero Downtime Migration (ZDM) that customers have been using for decades to upgrade their applications, database schema formats, file and table locations (or indices), operating systems, or perform a hardware refresh. In this article, we describe how a data replication engine can be changed without taking down either applications or the backup database.
Part 1 briefly discusses redundant systems, changing the replication engine, interoperating versions, and version independence.
Article (Part 1):
Swapping Replication Engines with Zero Downtime — Part 1
Part 2 discusses how the Version Independence approach (introduced in Part 1 of this series) can be improved using one or two additional nodes. It also explains how to protect the old production and old standby system(s) while the replication engines are being upgraded or switched.
Article (Part 2):
Switching Replication Engines with Zero Downtime — Part 2
Part 3 discusses various methods used to switch replication engines while avoiding data corruption or data oscillation (solving the so-called jagged edge problem).
Article (Part 3):
Switching Replication Engines with Zero Downtime — Part 3