SD-WANS: not the silver bullets they're cracked up to be?
via Flickr © one individual (CC BY-SA 2.0)
- SD-WANS are one of the big telco stories of the past 12 months, but they have their critics
- Never mind the band width, what about the band quality
- Are the customers waiting for an AWS of networking?
In the last 12 months or so we’ve heard much of the telcos’ software defined Wide Area Network (WAN) or SD-WAN. Like all ‘transformational’ projects, this one hinges on the idea of building a network by separating data and control planes (abstracting network and transport from the applications) and then orchestrating the resulting corporate network from the cloud. The approach seems to appeal greatly to telcos who will offer customers a hybrid corporate network solution which integrates different networks and technologies into a manageable, integrated whole.
It all sounds reasonable enough but according to some critics, the SD-WAN approach - as currently formulated - is flawed from both the business model and technical perspectives.
One of those critics is Gur Shatz, formerly the CEO of security specialist, Incapsula, and now CTO of Cato Networks. Yes, Gur has a dog in this fight - Cato is behind the Cato Cloud Network, described as a globally distributed network of Points of Presence, smartened up to provide an affordable, SLA-backed, global enterprise network backbone. So Cato is servicing the corporate network market the telcos are going after with their SD-WAN offerings and is also keen to partner with service providers to tap it.
Gur argues that telcos are perhaps not the right sort of organisations to be going down this path “Telcos just generally have this integration mindset - trying to build on top of things and not cannibalise their existing products and services. It probably needs a new set of players without baggage and they have to have a new product, not an integration around several different technologies,” he claims.
What providers really want
The current enthusiasm for SD-WANs, he says, is driven by the fact that they know they can’t do MPLS ‘pure’ anymore. The economics are just against it.
“They can’t keep on telling customers that as the bandwidth increases, they should increase their MPLS networks to compensate,” he says.
Such an approach is bound to fail because “they know that the basic technology is too rigid. They can’t ever make MPLS as cheap as internet, so they’re thinking, ‘OK, what we’ll do is find a way to prioritise so the critical information goes through MPLS and we’ll take all the growth to the internet side'.”
The SD-WAN solution enables the customer to keep the existing MPLS to meet the needs of priority applications while sating demand for extra bandwidth via new Internet capacity.
“If you want to increase capacity you need to find a way to do it and still be smart about what traffic goes through this line and what traffic goes through that line,” he says.
“So you can have a much more stringent SLA for MPLS than for the Internet and then you can just take some of the backhaul and extra traffic and get it to the cloud and it’s OK.”
But what if it turns out that the Internet-based service is just as good from an uptime, security and latency point of view. Won’t the customer then just move over to cloud-managed Internet VPNs for all services?
“That’s the risk. Once you put in a machine that directs some traffic across MPLS and some across the internet, then someone is going to turn the dial all the way round to the internet… and then it might shut down your business.”
However, that’s not a scenario that Gur thinks is going to happen any time soon.
Internet not ready… yet!
“While the Internet has improved in many areas - less congestion, increased capacity, increased competition and cheap bandwidth - the internet routing has not improved at all over the last two decades, so it’s not such a big risk at the moment. But if the internet’s routing does improve in the future, then at that point the dial is going to turn very very quickly.”
In the meantime, explains Gur, the industry needs a technology approach that that can gradually take over the MPLS role, but do it much more cheaply and easily.
“SD-WAN solution (as currently formulated) is too narrowly focused. We need a next generation that goes beyond load balancing and smart manoeuvring between MPLS and Internet - the next generation has to go after the backbone and make the internet on par with MPLS. There are are several companies that are expanding beyond the narrow scope of the SD-WAN and new categories will emerge,” he claims.
“Enterprises are really looking for the next big thing coming through. They’re looking for a new network category; something like the AWS of networking. It has to be a tightly integrated, built from scratch product that simplifies the whole process of building the network. As much as it has happened with the computer migrating to the cloud, the network smarts will migrate to join it.”