Extreme Availability with MIMIX Active-Active

Gone are the days of relying only on tape backup to recover your business on an IBM i server. Back in the day, restoring a server from tape could easily take a day or two, even for a small system, and any data created since the last backup was lost. Suffering down-time is inconvenient and having to re-enter the transactions since the last backup from paper records, even more so.

Today, IT teams are faced with challenges such as global business and partners who need access to business systems 24×7, all while complying with strict data laws and regulations. If your systems are not online, your customers can easily do business with someone else, and worse, may never come back. In the 21st century’s competitive economy, the impact of downtime can be irreparable.

Perhaps you have implemented a high availability and disaster recovery solution that replicates data from a single production server to a backup server that stands ready to recover operations in the event of failure or disaster. If, however, tolerance for downtime in your business has now moved towards single-digit minutes or even seconds, you may need to move beyond conventional thinking. You need to assess whether your solution can restore operations quickly enough to meet the requirements of your customers, business partners and staff.

There are Software-based high availability solutions for IBM i that go beyond the traditional expectations of recovery in tens of minutes. They can replicate data between two or more active servers, each of which can run production workloads simultaneously. This provides both workload balance across servers and HA/DR redundancy. This approach is known as “active-active” replication. No “recovery”, as such, is required to restore business operations. When a server crashes or needs to be taken offline for maintenance, you can redirect users to another active server, which is fully in sync with the downed server, by simply switching IP addresses. The result is recovery time that can be measured in seconds.

Having two or more production servers that are capable of reading and writing the same business data concurrently requires a change to the way the database is accessed. Your high availability solution must shift from using positional replication, where row numbers are used to retrieve and write records, to keyed replication, where the database retrieves records based on a unique key pre-configured in the database.

Keyed Replication Diagram

A collision resolution strategy will also be required to address those events that may occasionally occur when two servers are actively reading and writing the same data, such as when the data in a row is updated simultaneously in multiple locations. A holistic strategy also includes application optimisation to avoid collisions and the use of collision resolution methods to handle those that do occur.

It is important that your high availability solution provides methods to resolve conflicts and that your vendor works closely with you during an active-active implementation. Vision Solutions’ MIMIX Availability has supported Keyed Replication for active-active environments since the 1990s, allowing its customers to achieve unsurpassed high availability. Please feel free to contact Mynah Bird IT, who are the UK’s leading MIMIX reseller, to learn more about active-active replication, keyed replication and the rich set of collision resolution methods available in MIMIX.

Leave a comment