The Caringo Swarm system is a mature object storage system for private and public cloud usage and as a target archive system. Originally called CAStor, Swarm is delivered as software that is installed on X86 servers with attached storage. The scale out architecture allows the configuration to dynamically scale very large with commensurate performance. Caringo has implemented a symmetric parallel architecture that operates to manage the nodes as a distributed cluster. Nodes are automatically discovered and rebalancing occurs after booting from a network. All executing software is maintained in RAM on each node to enhance performance.
The object ID in Swarm is called the UUID (Universally Unique IDentifier) and consists of a unique generated random number. The UUID and a hash of the data is used for integrity verification. The hash algorithm may be replaced at a future time as algorithms evolve without affecting the data. Objects are stored as block data on the storage attached to the individual nodes. System and user metadata is stored with the object.
The Caringo system includes SwarmNFS to provide file access to the underlying object storage using NFSv4. Direct object access is accomplished with the S3 API, and a superset of S3 called Simple Content Storage Protocol, which conforms to HTTP1.1 and an HDFS connector. Geographic data distribution and protection is standard through sub-clusters on the same latency network. For remote clusters, Feeds automates the routing of objects for data distribution and data protection. Feeds are latency tolerant and take advantage of the full cluster-to-cluster capability of the system.
This Caringo Swarm Product Brief includes: