We are happy to announce a new design for the analytics dashboard and the ability to move your cache from one cluster to another in emergencies.
In August, we had the biggest incident in MemCachier history which affected two clusters on Amazon in the US-East-1 region simultaneously. During our post-mortem analysis we realized that, in extreme cases like these, a significant number of customers would prefer to move to a different cluster even if it means losing the contents of their cache.
As part of a wide range of improvement to better handle such extreme outages again, we’ve built out and decided to expose to customers the ability to move their cache to a different cluster.
The feature can be found in the redesigned analytics dashboard and is available to all paying customers. When used, the cache will be moved to the least loaded cluster in the same region and data in the old cluster will be flushed. The transition is seamless, meaning the configuration of the memcache client does not change. However, the DNS record change might take up to 3 minutes to propagate. It is important to note that this feature is meant to be used only in emergencies! Please always refer to http://status.memcachier.com to see if the cluster you are on is experiencing problems.
We are also changing how we create DNS records for customers as part of this change. While you used to be able to tell what cluster you were on from the DNS records, now the analytics dashboard will display this information.
One other use of this feature other than as a last resort in downtime, is for creating a second cache and ensuring your two caches are on independent clusters. If the two created caches happen to be on the same cluster, simply move one!
Rest assured that we’ve improved and continue to improve our ability to handle extreme outages without your involvement or loss of any cached data. But, it’s always nice to have a last resort.