Over the past few years, leading WAN optimization controllers have perfected their core optimization technologies such as protocol optimization, deduplication, caching and data compression. Today’s products aim to do more and eke out even better performance from enterprise networks.
The hot area now is application-specific optimization. This technology carries some risks, but with a little preparation, network managers can successfully use it to boost WAN performance. Here are some pointers.
1. Steer clear of Windows File Sharing.
Windows File Sharing (sometimes called Common Internet File System or Server Message Block file sharing) is a key application for many WANs and cries out for good optimization. Unfortunately, Windows File Sharing is one of the most dangerous areas to accelerate. Techniques that yield significant benefits, such as prefetching files or short-circuiting the inefficiencies of Windows Explorer, come with security and data integrity risks. As Windows has grown more sophisticated, so has the file system, its security model and application interfaces.
Instead, IT managers should look for less complicated protocols to optimize. If possible, consider replacing Windows file shares with SharePoint sites to eliminate the dangers of accelerating Windows File Sharing. SharePoint provides similar functionality for many user scenarios. Swapping out the protocol offers other advantages as well, such as a simpler security model and easier delivery of data to tablet and smartphone users.
2. Prepopulate optimization caches.
Often underappreciated, prepopulating caches can improve the user experience with almost no effort. The idea behind this is to guess what documents, files or web pages users will want the next day, and request them ahead of time during off hours. This speeds access when a branch office user requests data that’s in the cache.
Many IT managers view the practice as a waste of bandwidth, but it’s easy to do. Just launch a web crawler against the first pages of the Intranet each evening, for example, or copy the most recently accessed or created files off common file shares.
Because most organizations pay for network bandwidth 24/7 and make heavy use for less than half of that time, “wasting” bandwidth by copying files or web pages may not be a waste at all if users receive better performance the next morning. Some WAN optimization products even include a simple scripting language so that they can prepopulate themselves.
3. Reduce HTTP latency.
The mass of HTTP traffic running over application WANs has created a huge target for optimization. Basic caching offers some advantages, but the weaknesses of web protocols offer an opportunity for improvement.
WAN optimization products can speed performance by pre-establishing HTTPS connections across the WAN, avoiding the delays associated with establishing dozens of short-lived secure connections with many web pages and web applications.
For many HTTP-based protocols, the major performance killer isn’t the amount of data transmitted, but the effects of WAN latency on HTTP transactions. Reducing latency and round trips can have a noticeable effect on response time. When optimization products offer this feature, use it — and where they don’t, ask manufacturers to add it.
4. Tap traffic management.
Cloud computing presents some challenges for IT managers in delivering consistent response time to users. In branch locations where Internet connectivity is fast and reliable, employ the traffic management features of WAN optimizers to prioritize business-critical cloud applications over recreational and nonbusiness Internet applications, such as YouTube and music streaming services.
Where branch office Internet quality is lower, such as in international or very small aperture terminal-connected (VSAT) environments, WAN optimization still offers advantages. IT managers can backhaul their cloud application traffic over a highly optimized WAN toward their own regional data centers, and deliver it to the Internet from better-connected sites. The penalty of the additional hop through the data center is small compared with the advantages of optimizing typically inefficient cloud-based web applications.