High availability and disaster recovery are issues that can end up making a critical difference in terms of millions of dollars for certain companies. For example, on Black Friday if a busy website were to go down for even a few seconds it could mean dramatic losses.
In a recent DBTA webcast, Anil Kumar, Couchbase, went over issues related to high availability and disaster recovery, best practices to embrace and pitfalls to avoid in order to enable high availability for applications with NoSQL databases.
Being able to provide data redundancy - a copy of your data on a different data center - is a necessity. Single node type is the foundation for high availability architecture, he said.
“This is important because if users have a single-node architecture, they can easily grow their cluster,” explained Kumar. Intra-cluster replication is the process of replicating data on multiple servers within a cluster in order to provide data redundancy. This provides data for the user in case a server were to go down.
It is possible to create up to three copies of your data within a cluster. Failover in Couchbase servers automatically switches over to replicas, and there are constraints in place to avoid false positives. The failover will only occur one node at a time, but the DBA will be alerted to the other failing of nodes.
Another instance of providing availability of a user’s data is Rack Zone Availability. Rack Zone availability is placing groups of servers on physically separate racks. All together a cluster would contain three separate copies of a user’s data.
To view a replay of this webcast, titled “Enabling High Availability for Your Applications with NoSQL,” go here.