WAN optimization has emerged as one of the more vital, albeit confusing, technologies in the drive to centralize and consolidate everything from servers and storage to entire data centers.
Ever-shifting combinations of hardware, software and firmware make it difficult to determine how effective any one platform will be, or whether a single approach will work satisfactorily across all environments.
Most optimization products focus on caching necessary data on local appliances in branch offices or other remote sites, leaving the network more available for updated information. The latest improvements to that formula usually involve firmware upgrades, like the new version of Exinda that dumps packet-based caching for stream-based caching to improve capacity nearly 10-fold. The system also features auto-discovery tools that allow devices to communicate with each other without the need for centralized configuration.
But according to this report from Jack Germain, the focus is starting to shift to newer techniques like "object differencing" and application intelligence tools to deliver platforms with a more intuitive sense of what needs to be transferred at any given time. These developments involve shifting from the strategic methodologies of the past, designed to paste over applications' limited WAN abilities, to more tactical approaches in the drive to reduce traffic loads.
We could go one better, according to this blog from Steve Taylor and Jim Metzler. How about re-engineering the applications themselves so they perform better over the WAN? Sure, the ideal environment is a high-speed, low-latency network, but since that doesn't exist outside the LAN, wouldn't it be more efficient to simply rework the software rather than try and boost the network with lots of expensive gadgets?
At the moment, however, there are some applications that simply do not lend themselves to the WAN. Autodesk is one of the prime examples, due to its penchant for scrambling all the bites of a given file, which makes it difficult for deduplication tools in particular to determine what has been changed and therefore needs to be transferred.
Even though WAN optimization technology itself is decades old, applying it to modern virtualized networks still takes a bit of doing. The only certainties in this market segment are that demand is likely to remain high and there is still a wide variety of techniques that have yet to be fully explored.