SGI Expands Capability of SAN File System

SGI announced Thursday that CXFS 2.2, its shared file system for storage area networks (SANs) can now manage up to 32 different client machines within one SAN. CXFS previously handled up to 16 client machines.

A new file system from SGI is designed to enable multiple computers running different operating systems access a single shared, 64-bit file system within a SAN.

This improvement is designed to increase the scalability of SGI storage solutions and increase productivity by eliminating file duplication and the time necessary to move large files over the network.

CXFS is designed to let multiple computers running different operating systems — Solaris, SGI IRIX, and Windows NT — access a single shared, 64-bit filesystem within a SAN. This allows systems in the same environment to simultaneously access the same file at local or near-local disk speeds. CXFS aims to provide scalability, throughput, availability, ease of management, and lower the cost of ownership for storing, accessing, and managing data in a high-performance, multiplatform SAN environment.

CXFS improves data workflow by providing customers the performance and scalability of a SAN with the file sharing and connectivity of network-attached storage, addressing two fundamental challenges in data management: fast access to information and seamless integration of multiple operating systems. Because it uses the available SAN infrastructure, CXFS can deliver excellent I/O performance and bandwidth.

Benefits include:

• 64-bit scalability that supports file sizes up to 9 million terabytes and filesystems up to 18 million terabytes
• Heterogeneous client support for Solaris, IRIX, and Windows NT now, and other versions of Unix and Windows environments in the future
• Optimized distributed buffering techniques that allow for fast performance
• High availability with automatic failure detection and recovery

SGI CXFS 32-client support will be available later this month. CXFS for Solaris starts at $4,200 and includes for up to two processors (although it is capable of scaling up to 108 processors).