Highly scalable performance. Massive storage capacity. Heterogeneous platform support. These are not usually the characteristics associated with technologies that have failed to win major market share, but theyre just a few of the positive attributes of one such slow bloomer—SAN file systems.
But it just might be that storage area network file system technology was ahead of its time, and it looks as if that time is nigh. Business drivers such as demand for ILM (information lifecycle management) and more affordable computing systems are combining to expand SAN file systems beyond their current niche.
As the name implies, SAN file system solutions create file systems across SAN resources, including disks and tapes, allowing multiple servers to concurrently read and write to the same data set. SAN file systems are used mostly in applications where extremely large files are being shared and high data throughput (in the range of hundreds of megabytes per second) is needed.
SAN file system technology—unlike technology based on common file-sharing protocols such as CIFS (Common Internet File System) and NFS (Network File System)—can use Fibre Channel as its storage connection, providing a performance edge.
SAN file system technology is already relatively expensive, however, and use of Fibre Channel will up the ante by requiring expensive switches, host bus adapters and management software. The emergence of iSCSI as a SAN interconnect will likely drive new SAN implementations, especially at smaller companies, and will likewise boost the use of SAN file systems. iSCSI is far slower than Fibre Channel, but its also far less expensive.
With CIFS and NFS, data access is bottlenecked at the file server or NAS (network-attached storage) head level. With a SAN file system, in contrast, data flows freely from the storage unit to the servers. Metadata controllers restrict client access to storage resources, but once access is granted, the clients are free to communicate directly with the storage resources.
SAN file systems also establish a common file format to which all major operating system platforms can write. This is something you dont get with Windows-centric CIFS or Linux/Unix-centric NFS protocols, and it is a major reason for SAN file systems implementation at sites where Unix, Linux and Windows users need to collaborate on projects.
SAN file systems also give IT managers a single place to manage file and directory rights, which should make it much easier to secure and manage data.
Organizations interested in implementing an ILM solution will find SAN file system technology especially compelling: SAN file system products can migrate data seamlessly to various storage tiers without having to inform users and applications about where the data was moved.
Some SAN file systems, including IBMs TotalStorage, also can be configured so that rarely accessed data is automatically moved to another storage level—such as from expensive high-end arrays to tapes.
SAN file systems will also be a logical fit for emerging object-based storage technologies, where applications are able to indicate to storage devices their performance and data-retention needs.
Senior Analyst Henry Balta-zar can be reached at [email protected].