How to WAN Optimization

Organizations run their applications from central locations over the WAN, performance is impacted. Bandwidth becomes limited and consumption increases. Latency then occurs and there is contention. Elaborating on the above, that typically in an organization, the WAN tends to slow applications mainly because of its inability to understand the applications and WAN performance.

It does not help that there is more and more rich content across the WAN. Add protocol chattiness to the list and there is havoc on the WAN. In addition, contention impacts performance because low priority applications compete with critical applications for WAN resources. More over, lack of manageability occurs because how the applications and the WAN are performing cannot be seen.

Key business initiatives of WAN Optimization are compression and caching, acceleration, application control and visibility. Each of these areas cut costs and improves productivity. They also increase access to shared resources, improve disaster recovery, enable the roll-out of new applications, reduce time to trouble-shoot, automate remote device deployment and finally, manage remote sites with no local staff. On top of this, user productivity increases, money-saving applications like VoIP can be enabled and critical applications get prioritized.

Compression and Sequence Caching
Because organizations run applications with different traffic characteristics, IT needs a combination of reduction tools to increase WAN capacity. A next generation compression algorithm that runs in memory and uses flags to replace repeated data patterns. It provides anywhere from 2x to 10x the available WAN capacity.

It also provides dramatically better results than router-based compression, without adding the latency that those older technologies do. An improvement on web or file caching, it uses hard disks to store patterns, so it can store longer patterns and can store them for a much longer period of time than MSR can.

So it can eliminate repeated patterns seen days or weeks earlier, providing 10x to 100x increases in the available WAN capacity. But better than web or file caching, sequence caching operates on data patterns outside the construct of a file web page structure, so it can eliminate repetition even on modified files.

Application Acceleration
It is important to note that bandwidth is only one reason that applications perform slowly across the WAN. Latency is another major factor – sometimes overlooked and often not well understood. If a user’s WAN pipe isn’t full, yet users continue to complain about application performance, chances are latency is the fault.

Latency can affect applications at different layers – sometimes it’s TCP’s behavior that slows an application, as in FTP and other bulk transfer applications, and sometimes it’s the application’s own Layer 7 protocol. The best example of when the Layer 7 protocol is the reason for the slow down is Exchange using MAPI, file access on Microsoft file services using CIFS, and web traffic running over HTTP.

For those applications, accelerating TCP will do nothing to speed up these applications because they send data in even smaller chunks than TCP does, so opening up the TCP window size doesn’t help them until they’re accelerated at Layer 7. Then TCP acceleration can kick in as well to speed the application. PFA accelerates applications at the TCP layer, and AppFlow accelerates Exchange, file services, and web traffic.

TCP provides the broadest, general acceleration, and the application specific acceleration is essential for those business critical tools and to enable server centralization and/or consolidation. For those applications, once you accelerate them at L7, then the TCP acceleration also kicks in to further speed their performance. CIFS, the protocol for file transfers, and MAPI the protocol for email transfer, require thousands of round trips to complete a transaction such as opening a file or sending an email attachment.

The impact of latency on these round trips can increase the file transfer time from seconds on the LAN to many minutes over the WAN. The WX has specific acceleration capabilities for CIFS and MAPI and can overcome the effects of latency. The WX acknowledges data requests locally and requests all data blocks for a file with out waiting for acknowledgements to transit the WAN. It then pipelines the data across the WAN, accelerating file transfers and overcoming the effects of latency.

Application Control
Provide more bandwidth and accelerate applications to overcome latency, you still need to control which applications should gain access to the precious WAN resources. Terms of applications and set policies for which applications should get first access to the WAN via very simple templates and wizard-based tools.

If implementing VoIP, for example, there’s a built-in template that makes setting up QoS for VoIP very easy to click and apply. One key attribute of the QoS capability is its on-going understanding of the available bandwidth. This feature is key to understanding how much to throttle back certain applications. It also enables better performance over shared networks with variable bandwidth, such as MPLS.

Visibility and Control
Visibility, with all its critical reporting functions, provides IT with WAN, link, and most importantly application performance metrics. The tools are critical for troubleshooting as well as future planning. Understanding these four key technology areas is vital if you want your wide area network to provide key value to your enterprise as well different sources of ROI.

Reference: Asian Channels