vCE? What is a virtual customer edge?

vCE? What is a virtual customer edge?

If you’re looking at an SD-WAN solution, some of the providers will offer you a vCE (virtualised Customer Edge) as part of the deployment. Are there advantages, and should you take this option if offered?

If you’re in a site where the bandwidth to the location is large and relatively low cost (i.e. 1Gb/s connections within a major city), then this can be a good option to look at, but if you are in a more constrained location (remember the Three Bottlenecks of WAN), then there might be some challenges…

A virtualised CE is NOT a customer premise equipment located on your site, but hosted by the service provider in their PoP (Point of Presence) location. It is an ideal option if the bandwidth available between your location and the service provider PoP is provided over high-bandwidth links, and was envisaged primarily to support locations that would be connected by fiber to the providers PoP.

This means that the devices that are on site are just termination and presentation matching between the link provided and the on-site network equipment. This scenario works really well when the vCE and the on-site switch for the WAN access layer, with the on-site switch providing on-site Layer 3 routing, avoiding the need to send data down the link to the provider and back (which would increase latency) for local site routing.

If you have are looking at supporting data compression and acceleration proxies on the CE, this many not be the best choice for you for a couple of reasons.

If you are using data compression, and have a highly busy site, it is possible that the link between your location and the vCE becomes the bottleneck, reducing the effectiveness of any data compression gains by blocking the link to the site (if the site is receiving traffic), or from the site (if the site is sourcing traffic).

And similarly if you are looking at minimising latency through the use of proxies or acceleration technologies on this device, consider that the latency is considerably increased over using a local device already simply due to the geographic separation of the your location and the provider vCE location, negating some of the benefits you may get if the same function was located on-site.

Conclusion

If you have no need of these technologies now, and have a high bandwidth capability to the provider PoP (more so than the amount of data that you transact with other sites), then these are good solutions for you, and minimise the provider footprint on your site.

If you need to use NFV later on and that requires an appliance on-site, then you can always have this added by your provider at a later stage (make sure your contract with them allows you to move from a vCE to a virtualised Customer Premise Equipment solution at minimal cost)

If you can’t get access to high bandwidth links in your location, or have need of the NFV functions right now, then a vCPE solution should be the route you choose.

John Dixon

John Dixon is the Principal Consultant of thirteen-ten nanometre networks Ltd, based in Wiltshire, United Kingdom. He has a wide range of experience, (including, but not limited to) operating, designing and optimizing systems and networks for customers from global to domestic in scale. He has worked with many international brands to implement both data centres and wide-area networks across a range of industries. He is currently supporting a major SD-WAN vendor on the implementation of an environment supporting a major global fast-food chain.

Comments are closed.