A new class of products enables file caching over WANs, making it possible to access data over long distances at LAN speeds. Storage-caching devices eliminate the delays that typically plague remotefile sharing.
In today's office environments, people move files between dispersed locations via e-mail, FTP or by physically shipping tapes overnight. These data-sharing methods lead to delays and result in many unsynchronized copies. Furthermore, remote file-sharing standards such as Common Internet File System (CIFS) and Network File System (NFS), which originally were developed for LANs, cannot be used over wide areas because they are synchronous and chatty. These protocols use many remote procedure calls (RPC), which introduce delays that can cripple file operations over long distances.
Storage caching moves beyond Web caching, which uses HTTP to distribute read-only content. Storage caching gives companies a practical way to extend complex file protocols such as CIFS and NFS over WANs with full read/write capabilities.
At the heart of storage caching are appliances equipped with software that allows files to be shared between a central storage pool and remote sites. A cache appliance attaches directly to a LAN at a remote site and to a cache server in a data center via a WAN.
These appliances communicate using a distributed, fault-tolerant storage-caching WAN protocol. Protocol translators in each storage-caching device convert NFS or CIFS traffic for WAN transport and back again. To LAN users, a cache appears to be a network-attached storage (NAS) device. Users then can access shared files through standard file-sharing protocols.
When a remote user attempts to access a file, the request travels over the WAN to fetch the file from the cache server. Streaming techniques improve access times even on the first file fetch. Subsequent file requests at the cache translate into cache hits - those that can be served locally at LAN speeds.
As users modify files at the cache, the files are written back to the cache server over time, which vastly improves response times for user applications at the remote site. The modified files are efficiently synchronized with the corresponding ones on cache servers using smart file-differencing techniques. Then when any other remote site requests a copy of the file, the cache server delivers the most recent update.
The technology provides capabilities such as streaming, compression, encryption and file sharing.
Streaming techniques eliminate most of the message-passing delays that slow file transfers. They send groups of file-sharing transactions that would otherwise need to wait in line for responses one at a time and are the key to countering latency penalties.
Bidirectional distributed differencing minimizes bandwidth-monopolizing file transfers by transmitting only the differences between file versions. It intercepts the data before it is transferred over the WAN and extracts the changes. The changes then are sent over the WAN and applied to the file on the other end.
Storage caching is a vast improvement for system administrators, allowing them to manage and monitor storage systems from a central storage pool. It simply plugs into existing storage, hardware and network infrastructure to tame wide-area file-management performance despite inherent WAN latency and reliability issues.
The most obvious benefit of storage caching is storage consolidation. By allowing fast access to files at a central location, storage caching effectively removes the need to distribute NAS devices or file servers among remote offices, reducing capital expenditure and management costs for these devices.
Sadalgi is a software engineer at Tacit Networks Inc. He can be reached at firstname.lastname@example.org.