Today, the Internet and dedicated wide area network (WAN) resources allow direct connections between the data center and remote offices. Remote offices can now access applications and data from the data center, and backup tasks can take place without the intervention of any remote employees.
Wide-area file services (WAFS) technology allows remote users to access files at LAN speeds over a WAN. This alleviates the need for remote storage so corporations can consolidate servers and storage back to the data center.
Protocol acceleration deals with the "chattiness" of applications. Protocol acceleration reduces the number of handshakes required as applications open or close files and boosts data movement by altering the size and characteristics of network packets.
Data compression reduces the sheer amount of data that must be transferred between points by replacing repeating data patterns with smaller "tokens." That data is then reconstructed at the other end of the link.
Cache is simply local storage, allowing data to be stored temporarily at the remote office rather than moving it repeatedly between points.
WAFS is a generic term that indicates basic file access acceleration and some degree of WAN optimization. However, vendors use different names for the technology. Riverbed Technology Inc. uses the term wide-area data services (WDS or WADS) to include WAFS along with a richer set of WAN optimization and file access features. Cisco Systems Inc. uses wide-area application services (WAAS), which is also more comprehensive and feature rich than WAFS. Brocade Communications Systems Inc. uses the term file area network (FAN) to refer to a collection of remote file management and access technologies. The term wide-area data management (WADM) has also recently appeared.
Role of cache in WAFS
Cache serves two roles in a WAFS platform: reducing WAN congestion and overcoming WAN disruptions.
Cache reduces WAN congestion by easing redundant data transfers. For example, each time an employee in a remote office opens an important file, that file demands bandwidth as it is transferred from the data center. If a frequently used file is stored in the cache of a remote WAFS appliance, it can be accessed without using WAN bandwidth. Any changes to the file can then be synchronized back to the data center as time and bandwidth permit. Note: The initial file transfer or preloading can take significant time across the WAN.
Cache also overcomes the problem of WAN disruptions. No WAN is 100% reliable, and disruptions invariably occur. Storing frequently used files in the local appliance allows users to continue using those files even if the WAN connection fails. File synchronization can resume once WAN connectivity is restored.
The amount of drive cache depends on the size of the appliance. A WAFS appliance intended for small offices may only use 100 GB with up to 2 Mbps of WAN connectivity. Appliances for larger offices may reach 750 GB with 10 Mbps of connectivity, and enterprise-class WAFS appliances may supply 3 TB with more than 300 Mbps of connectivity.
WAN optimization is an important attribute of WAFS technology. The amount of optimization varies by application. Some WAFS appliances provide TCP acceleration by repackaging the data into larger packets, which improves the WAN performance of a TCP application. The latency of specific applications can be further reduced by reducing the number of individual handshakes needed to transfer data. Storage administrators should understand the applications in the enterprise and consider the amount of acceleration that each application will achieve.
WAFS file limitations
WAFS still has limitations. One is that WAFS relies on caching to store key files locally. This means there may be several copies of a file located in the network. Not only can this compromise security by multiplying the exposure of a file, but it also makes it difficult to keep versions straight, as different users access files from different locations.
Furthermore, because WAFS does not support differential changes, file synchronization is not very efficient. When a file changes, even a small change, the entire file must be sent again. As a result, active files may demand significant bandwidth as they try to remain synchronized between locations. The way to mitigate these problems is to discuss these limitations with your WAFS vendor and establish best practices.
This was first published in January 2008