Automated Database Failover

Get our Newsletter

In today’s data-intensive business environment, every minute counts. Enterprises experiencing unplanned downtime can suffer an average $600,000 per hour in lost revenue.
Many companies solve business downtime via load balancing and high availability failover solutions. However, these existing HA solutions still do not guarantee application uptime upon a database server failure as the application Moreover, they are costly and complex to manage. For example, in MySQL deployments, there lacks a truly robust HA solution in the market.
Therefore, database availability is still an issue for customers with critical applications. Customers have resorted to are alternative solutions, none of which are optimal:
1. Up to 60% of companies have delayed security patches in order to reduce downtime, which exposed them to regulatory and compliance issues.
2. Some customers have implemented a manual failover scheme requiring an application restart. Upon a database outage, application downtime could last from 20 minutes to 3 hours, resulting in a loss of potential revenue for the Enterprise.

The Heimdall’s automated database failover solution avoids the above scenarios for maximum uptime. The key benefits of the Heimdall approach are the following:

1. Persisted Connection Failover: Most solutions today (e.g. F5 Networks, AWS Elastic Load balancer) are at the TCP/DNS layer, resulting in connection timeouts and reconnects. Upon a detected database failure, the Heimdall Proxy queues up front end connections and gracefully connects to the standby database instance, reducing application errors and exceptions. Failover is transparent to the end user.
2. No application or database code changes: Scaling out the database tier always requires rewriting the application; not anymore with Heimdall Data.
3. SQL-aware read/write split for complex routing to different geographies or data sources.

Automated Failover Technical Brief