The Dark Side of Disaster Recovery

For decades, disaster recovery dominated the IT landscape as the undisputed architecture for business continuity. But IT has woken up to the ineffectiveness of this approach. All your compute resources sit idle until a disaster strikes, and when it does, all your operations go offline and recover “sometime soon.” This approach no longer serves, in the age where every business is expected to remain accessible round the clock. We live today in an era of Always On service, where there is zero tolerance for downtime. Many online businesses have reported huge losses to revenue and significant damage to reputation when their online services went down. Today, enterprises need to architect their infrastructure for continuous availability rather than investing time and effort in building disaster recovery structures.

The New Gold Standard – Active/Active Data Center

With the advent of virtualization, big data, cloud technology and modern database management systems, active/active architectures are no longer an option but a necessity for ensuring business continuity. But they present a new set of challenges as they are costly to implement, complicated to build, require extra equipment, and demand low latency. These architectures have proven especially hard to build at the data tier. Fortunately, database load balancing software makes it much easier and affordable to architect for active/active database operations. The traffic management capabilities of a load balancer enable enterprises to cut costs and reduce downtime by leveraging the following four features:

1. Automatic support for read/write split. The first step to active/active operations is having multiple servers to support database reads. The challenge is that applications must be recoded to send reads to active secondaries. Database load balancing software solves this problem, by automatically directing write transactions to the primary server and reads to the secondary servers. Automatic read/write split delivers instant app performance and reliability gains without needing app modifications.  
2. Auto Failover. A load balancer leverages the failover mechanism built into modern databases to enable app-transparent failover. The database performs the replication and the failover – the database load balancing software holds inbound requests during the failover so apps see delay but not errors.
3. Geo-enabled Load Balancing. Database load balancing software constantly measures time to first byte to direct traffic to the fastest-performing servers. It also balances the read load across all available secondaries, delivering fast application performance.
4. Replication-aware Monitoring. Database load balancing software is designed to consistently monitor replication lag and identify nodes that have fallen behind the predetermined thresholds. It then avoids sending requests to out-of-date servers, so it doesn’t serve stale data.

Active/active operations are costlier as compared to disaster recovery, but only by about 20% while delivering 35% more capacity and enabling non-stop operations. This increase in cost is outweighed by the increase in capacity, improved uptime, enhanced performance and optimum asset utilization. When continuously available architectures deploy a database load balancer, they also enable zero-downtime maintenance. This new path-breaking technology is empowering enterprises to avoid disruption and ensure seamless operations by moving from disaster recovery to continuous availability.

Database load balancing software is the most reliable and affordable way to architecting active/active data centers. Without needing any code changes, a load balancer delivers dramatic performance improvements even in the most demanding situations. With features like in-memory query caching and app-transparent scalability, it delivers the most value to online businesses whether they are operating on-premise or in the cloud.