High Availability

Take advantage of the high availability features built into Gear6 Web Cache. Never lose revenue because of cache tier failures.

Typical Memcached deployments do not comprehensively address web site requirements for high availability. Depending on your web architecture, a single failure can disable your entire web caching tier. Starting a new cache service to replace a failed one can take minutes, and restoring cache data until performance is again adequate can take hours.This presents serious obstacles to scaling a large site, since you have to over-provision for failure events, and may feel the need to implement complex cache high availability into your applications. High availability is part of the foundation of Gear6 Web Cache. It is based on apowerful cluster architecturewith multiplefailover configuration possibilities.

Gear6 Web Cache can operate in two high availability modes:

Continuous Service Availability through Clustered Architecture

With continuous service availability, instance 2 is available immediately after a failure, but most be reloaded with cache data.

Standard Memcached services are typically deployed across multiple hardware modules, but Memcached behavior is unpredictable if a module fails or some other failure occurs in the service. Instance IP addresses can vanish, triggering excessive latency or cascading failures.

Gear6 continuous service availability uses a clustering architecture that automatically fails over a failed instance to another available module. It keeps instance IP addresses alive and Memcached services functioning.

With continuous service availability, the service, but not the cache data, is maintained in the event of a Memcached failure. This is the lowest-cost, highest-density way to deploy Gear6 Web Cache. It is suitable for applications where additional Memcached latency is acceptable after a failure has occurred, while new modules are loaded with cache data.

Gear6 Web Cache can also be configured for continuous data availability, so that normal operations continue without performance impacts throughout a failure incident. Learn more about continuous data availability.

Continuous Data Availability through Replication

With continuous data availability, instance 2 is immediately available, with all its cache data, on a different module after a module has failed.

The continuous service availability features of Gear6 Web Cache automatically and transparently fails over to a new module when a module failure occurs, and normal operations resume as the new module is loaded with cache data.

But for some applications, cache performance may be unacceptably reduced as the rebuilding takes place. For this reason Gear6 Web Cache can also be deployed with replication, providing continuous data availability. With replication, cache data is always duplicated in another location, and when a module fails, the Memcached instance on the failed module is immediately replaced with another one already containing the same data. Memcached operations continue without interruption or slowdown. Spikes in database and application load are avoided.

Replication typically requires twice as much memory but is often justified for large and mission-critical applications. And because of the Gear6 Web Cache hybrid DRAM-flash architecture, a replicated Gear6 solution is typically still much less costly to purchase and operate than alternative Memcached solutions.