Peer Fusion Data Storage Solutions

The Peer Fusion cluster technology can be configured for all storage needs:

• Enterprise Requirements:



• Productivity.

The Peer Fusion File System (PFFS®) increases productivity during the software development cycle and the operational cycle. Software developed using standard POSIX filesystem semantics can run in the production environment without changes by using the Peer Fusion resilient parallel filesystem. The entire hierarchy of the Supra-Clusters™ is accessible through standard operating system utilities as a unified namespace. You developed your software using open/read/write/close/etc and the PFFS® provides the same filesystem calls in its large scale, distributed, and resilient cluster of clusters.



• Data Protection.

The protection of data is a fundamental aspect of the PFFS®. Data resiliency is implemented through erasure codes, so there is no replication of data. Data is encoded for resiliency as it is ingested so there is no intermediate step of replication pending encoding. The PFFS® supports several level of resiliency so you can decide which is best for your data protection and performance requirements.



• Cost Reduction.

There are three major areas of cost reduction in the PFFS®. The first is achieved through the elimination of data replication and the resulting significant reduction in the requirements for hardware, space and power. The second area of cost reduction is the simplicity of operating a PFFS® cluster which requires fewer trained personnel. The third major area is increased development efficiency resulting from the similarities between developing software in a POSIX filesystem environment and running the software with the PFFS® which is also a POSIX filesystem.



• Simplicity.

A paramount design objective of the PFFS® is simplicity for the IT personnel who operate the clusters. The cluster configuration files are very small and simple, with the cluster peers obtaining most configuration information initially from the gateways as part of network discovery and subsequent changes through heartbeats. A VFS interface is provided at three levels: the peer level, the cluster (i.e. gateway) level, and the Supra-Cluster™ level that is compatible with most utilities and system calls.



• Scalability.

Supra-Clusters™ have virtually no architectural scalability limits as they can aggregate thousands of independent PFFS® clusters for a total of tens of thousands of peers. Many Supra-Clusters™ gateways can be used to ensure performance as required. All the peers can process their locally stored data in parallel for the most efficient throughput.



• Performance.

The ingest performance of Supra-Clusters™ is very high as data received by the gateways is forwarded directly to the appropriate PFFS® cluster. All the cluster gateways and peers take maximum advantage of the hardware available for parallelism using threads. Network access is implemented using multiple threads per Network Interface Card (NIC), the process of encoding/decoding data for resiliency is divided among the CPU cores, pools of threads are used to respond to I/O requests, etc. Client applications running on each peer read their local data through the high-performance PFClient® filesystem, and can store the data they produce locally, or in their local cluster through a PFFS® gateway, or in any Peer Fusion cluster through a Supra-Cluster™ gateway.