WANdisco’s patented replication technology enables workload to be balanced across multiple active peer NameNodes that simultaneously support client requests and continually synchronize across any number of data centers any distance apart. Clients have LAN-speed read/write access to the same data at every location without downtime or data loss. Support for rolling upgrades eliminates downtime when new versions of Hadoop are deployed. These capabilities meet the most stringent regulatory and business requirements for data availability and disaster recovery.

With Non-Stop Hadoop, there are no passive standby servers. In WANdisco’s active/active architecture, all servers across all locations are fully readable and writeable. Money isn’t wasted on read-only backup servers and clusters that can’t be fully utilized. For even greater efficiency, features such as selective replication and asymmetric block replication allow different hardware footprints in each location.

Cluster zoning enables a ‘virtual cluster’ or zone within a cluster to be deployed that isolates the most resource intensive data load and in-memory applications to the highest spec servers, while less critical batch applications run on low-end commodity servers. This offers significant cost savings by eliminating the need to deploy high-end servers throughout a cluster, while maintaining quality of service for all users.

WANdisco’s active/active architecture enables data ingest at any number of locations simultaneously, automatically replicates it to any number of locations, and allows it to be analyzed anywhere without any single point of failure. This eliminates the administrator overhead and risk of error involved in using DistCp and other tools to copy data from its source to a central location for analysis. This capability is essential in large global organizations where data is generated from widely distributed sources and timeliness and accuracy are critical.

In large deployments where selective replication is used, data may not exist at the location where analysis is run. WANdisco overcomes this by enabling Hadoop to be data center aware, making global company-wide roll-up analysis possible regardless of how data is distributed across an organization.