Yes, WAN Optimization may be the greatest thing since Cat-5, but there is a significant drawback to the solutions on the market. The various optimization techniques that these appliances apply on the application packets all affect monitoring and reporting tools to varying degrees, and distort the results of performance reports.

This means three things for those of you who are deploying a WAN Optimization device: 1) you will lose visibility into important metrics, 2) it will be more difficult to troubleshoot network problems and identify bottlenecks, and 3) it will be problematic to even quantify the benefits of the WAN optimization devices.

Think about that. You could spend millions of dollars on WAN optimization equipment and not even know if you're getting any tangible return on investment. The simple question: "How much will this solution improve our network and application performance?" can't be answered with any accuracy.

One of the reasons is how WAN Optimization solutions handle packet header data. This is the data which passive network performance tools use to give you detailed information about end-user response times. WAN optimization devices use techniques that change the packet header data so that data it can be transferred more efficiently.

For example, Tunnel-based architecture (or "Layer-3") optimization solutions tunnel the client-server traffic though dedicated TCP ports between the WAN optimization devices. All the flows disappear into a single WAN Optimization port, so it can be difficult to determine what type of traffic you are dealing with.

LAN-WAN segmentation (or "Layer-4') optimization solutions use two peer devices on either side of the WAN that optimize the TCP connection between client and server by breaking a single TCP connection into three. The first goes from the client to the WAN Optimization Device (WOD) at the edge, the second goes between the WODs at the edge and at the core, and the third goes from the core-side WOD to the server.

But the problem with segmentation is that most monitoring systems will assume the existence of a single TCP connection and will therefore only report on a single TCP segment - typically the one between the core-side WOD and the server. Worse still, if you weren't aware of this, you might take a look at the performance monitoring tool thinking the network is lightning fast, when all you're getting is the local LAN speed - not the end-to-end connection - the only thing worse than being uninformed is being misinformed.

Well, I suppose having your toes gnawed off by weasels is worse than being uninformed. But not by much.
--------------
More Information:

Press Release
- NetQoS Integrates with Cisco WAAS to Deliver End-to-End Application Response Time Reporting for WAN Optimization

Webinar
- WAN Optimization and measuring the results

On WAN Optimization