High availability within a cluster

Last Updated : Apr 11, 2021 |
Note:

  • Context Store does not support high availability for single-node deployments. If the single node fails, you cannot recover the data.

  • For a cluster with more than one node, Context Store supports data preservation for only one node failure. For more information about multiple node failure and high availability, see Avaya Context Store Snap-in Developer Guide.

Context Store uses the Avaya Breeze® platform and data grid to achieve high availability. For High availability, the Context Store cluster must consist of two or more Avaya Breeze® platform nodes that are dedicated to Context Store. Context Store creates a data grid across the Avaya Breeze® platform servers in the cluster where you deploy Context Store, so that the servers can operate in an active/active mode.

To achieve high availability, Context Store data grid uses the following techniques:

  • Self healing: The Context Store data grid has self healing and disaster recovery capabilities that minimize the chances of a downtime. The data grid also preserves its states so that, in case of a system failure, you can recover the data grid in its latest state.

  • Data sharing: All the servers in a Context Store cluster shares the same data grid. Even if one server is unavailable, the other servers in the Context Store cluster can use the data present in the data grid.

  • Data partition and redundancy: Context Store partitions the data that is present in the data grid and distributes the data across different machines. This distributed structure enables the system to hold more data than the capacity of a single machine and also helps attain high availability. The system keeps a backup of data from each partition so that if one source becomes unavailable, Context Store can retrieve the same data from the backup location.

Server failure

The Context Store cluster is accessible through a cluster IP address that is maintained by a highly available load balancer. The load balancer distributes HTTP requests across each of the Avaya Breeze® platform servers in the cluster. If one of the servers in the Context Store cluster is unavailable, Context Store can still work without loss of data and significant impact to its performance. However, unavailability of two servers might result in loss of data and failing HTTP requests. If two servers are unavailable in a cluster, restart the cluster to clear old context data and restore Context Store to a known state. When the cluster is restored to a known state, data replication in the cluster is not accurate for updates that occured during the failover period. The data and EDM on the other cluster, which serviced the contexts during failover, is accurate.

Note:

You can enable load balancer only after adding nodes to your Context Store cluster.

If any of the servers in the cluster are unavailable, Context Store redirects the client requests to the primary partition. If the primary partition is unavailable, Avaya Breeze® platform determines a new primary partition to direct the client requests. Context Store might have an increased response time during this negotiation process, until a new primary partition is assigned.

When an unavailable server becomes available, Avaya Breeze® platform readjusts the partitions to accommodate the new server. If the server that becomes available contains a primary partition, Avaya Breeze® platform determines a new primary partition to serve the client requests. Context Store might have an increased response time during this negotiation process, until a new primary partition is assigned.