Cloud-based WAN Optimization

There is an unmistakable industry trend to offer every IT function in the cloud, if at all possible. WAN Optimization (WANO) is no exception. WANO is used to mitigate the effects of a WAN (low bandwidth, latency, and packet loss) to get improved application performance (fast end-user response time) and also to reduce the bandwidth consumption itself. WANO in its most popular form is symmetric in nature – an appliance called a WAN Optimization Controller (WOC) is placed at each end of a WAN connection (at the data center and at the branch).
WOCs achieve their goal using a combination of techniques- compression, de-duplication, caching, TCP optimization, and chattiness reduction at the application level. These techniques are applied across the whole segment of the WAN by the two WOCs. All the benefits of the WANO start right in the access segment or the last mile at both ends of the connection.

The issues with WOCs, particularly among small and medium businesses, are their initial cost, ongoing maintenance, and complexity of managing them in-house. CAPEX is always hard to come by for small/medium businesses. Managed Service Providers (MSPs) can alleviate this issue by offering a fully managed WANO service with charges on a monthly basis. The OPEX model offered by the MSPs can be called a “semi-cloud” solution from the customer point of view and appears to offer comparable benefits of fully owning and operating the WOCs without many headaches.

Another twist in this equation is that WOCs may not be as efficient over the public Internet as opposed to a somewhat expensive MPLS network. Small/medium businesses invariably use Internet-based VPNs to construct WANs connecting their widespread locations. So the MSP model may not always work satisfactorily – especially in cases of customers with locations across the globe.

In the last few years Cloud-based WAN Optimization (CWANO) providers have emerged on the horizon (a company called Aryaka is leading the way). CWANO vendors, in their basic service, avoid deploying any appliance at all on the customer premises. The magic starts and ends at their Points-of-Presence (POPs) in the cloud to which the customer locations connect. Sophisticated and carrier-grade WOCs are deployed in the cloud by the cloud provider. They also provide convenient web-based management tools to configure the service and monitor the WANO results achieved.

The OPEX model of pricing offered by CWANO providers is much cheaper than owning and managing the WOCs. Customers don’t need to buy expensive MPLS connections as the solution runs over the public Internet. Is it too good to be true?

Yes, a closer examination of the CWANO reveals that the optimization takes place only in the middle segment of the WAN journey – between the POPs. Nothing happens on either side on the last mile. The claim is that the middle segment which can span continents for international connections is often where the Internet performance is at the worst (high latency and high packet losses) and that is where optimization is needed. However, the end links also can significantly affect performance, possibly due to low bandwidth.

The best way to verify if “WANO in the middle” provided by cloud-based providers really helps is to conduct a proof-of-concept and compare with the baseline results. If we can also compare with the case of full WANO with WOCs at either end, then we can make an apples-to-apples comparison.

But the CWANO providers are also smart enough to hedge their bets. As an option (many times at no cost), they can provide some kind of an appliance at both ends and provide full segment WANO. In this case, they are not a true cloud-based provider but more like an inexpensive MSP.

In all fairness, the CWANO providers have enlisted many satisfied customers with global locations. If you are planning to use CWANO, as the saying goes, trust but verify.