Routers Are Facing Extinction: Let’s Not Save Them

Routers are everywhere. There are literally gazillions of them. They overpopulate every type of network, not to mention the old ones that are past their use-by date and being constantly being consigned to the scrapheap. I have a whole heap of these -- need one?

The fact is that routers roam across every network unchallenged -- they have thrived in a more traditional, primitive IT age. The dinosaurs of the connectivity world, if you will. And much like the dinosaurs, routers are facing extinction -- and I see no need to save them.

In fact the overpopulation of routers is at the very heart of their demise. Service providers have become so used to deploying them that they have almost forgotten what they're for. Most of the routers that live in networks today are massively underutilized. They are often placed as simple demarcation devices for basic aggregation, or, more embarrassingly, deployed as "Shadow Routers" when the primary router isn't man enough for simple monitoring functions.

Routing has always been unnecessary at the service edge: Consider a basestation connected by a single backhaul link, or a business with a single access line. What's the point of a router when there is only one possible "route" to take? It's like a traffic cop directing traffic to go straight on down a one-way street. It's a good "make work" project and might go unnoticed if service providers had money and time to spare. We all know they don't.

Fashionable lingo across the telecom industry is currently based on phrases and trends such as "streamlining," "flexibility," "agility," and the need to "drive out cost" and "radically enhance the delivery of new services." Today's virtualized network reality is built on the white box and will herald a new ice age for the router.

There are many ways to eliminate routers and nudge them along towards extinction. Virtualizing them is one popular strategy. Centralizing them is another. We prefer a kinder, more gentle approach. Like giving unwanted clothes to those in need, we started the first not-for-profit "Router Relocation Service." We try to find them homes where they will feel useful and become fully utilized again. We're sending them back to the data centers, central offices and other locations where they can actually route traffic and rebuild their self-esteem.

From what we have seen from our own direct experience, we believe that replacing routers with capable vCPE edge strategies can drive out 90% of the cost and complexity routers introduce. Routers are a lot like a boat-car -- versatile, but neither a great boat nor a great car. A proper vCPE strategy can, I believe, actively overcome this identity crisis plaguing routers and take great strides towards dramatically improving performance.

So that's it. The extinction of routers is likely to be less explosive and less immediately definitive than the fate of the dinosaurs, but it is no less inevitable. It's such a shame that such a once pivotal piece of network engineering seems destined to be phased out -- it almost deserves a more dramatic finale. I wonder if I could get a giant asteroid on eBay?

— Scott Sumner, VP solutions marketing, Accedian Networks

Page 1 / 2   >   >>
kq4ym 12/6/2015 | 4:31:47 PM
Re: Very droll, but.... Although the argument "What's the point of a router when there is only one possible "route" to take" is valid, one wonders how soon the extinction will actually be. The residential user is still going to be one of the last to disappear of course.
brooks7 11/20/2015 | 3:08:57 PM
Re: DSLAMs ? 250K DSLAM would have to be like 10K ports.


jabailo 11/19/2015 | 6:08:39 PM
Re: DSLAMs ?  

$250K switch?  Yeep!

[email protected] 11/19/2015 | 4:19:59 AM
Re: Very droll, but.... "router toss contest"

Are you the Johnny Knoxville of telecoms?

Sounds brilliant...
brooks7 11/18/2015 | 7:21:33 PM
Re: DSLAMs ? Dlams and olts are Ethernet switches. seven Ps stupid autocorrect.
[email protected] 11/18/2015 | 6:48:57 PM
Re: Very droll, but.... Regarding DSLAMs - I'm referring mostly to cell site and enterprise business services routers, which typically would connect into the access with direct fiber connections or via an Ethernet ring (G.8032v2, in the Colt use case), which don't involve routing, as most are layer 2 in the last mile. Of course MPLS services require routing, but typically static routes that are pre-defined, for the most part. Residential services is another animal, that may survive the router ice age :)
[email protected] 11/18/2015 | 6:46:05 PM
Re: Very droll, but.... Cost savings are enabled by the transition to virtualized NIDs, smart SFPs, and other edge alternatives that run in the 100-250$ range, compared to traditional NIDs or PE units that run closer to $2000+ ... this is backed by several key Tier-1 ROI case studies, including Colt's business services, where they just reported at the GEN15 show (where we are now), that they have realized more than 85% cost savings to date (over a 3 year rollout), and still expect to realize several more percent in the coming year. I'll try to get the slides and load them up here.

We had a router toss contest here in Dallas, with great participation from even the most unlikely parties - you'll see Michael Howard's form in recycling routers once we have the video rendered. Quite a moment.

Check out our extinction video here, and learn why we think we're entering a router-free era.
[email protected] 11/18/2015 | 3:33:41 PM
Re: Maybe droll - but also prescient I don't disagree with anything you say, so let's pin down what we might reasonably mean by extinction - and I'm not holding out for a 0% dinosaur scenario.

At what point does anyone think that 95% of current edge routers will be replaced with white boxes? That's a big percentage.

A technology refresh cycle? Hmmm, I think a few. And this is all presuming quite a lot about the performance of white box/VNF substitutes in real world production networks that are under increasing security threats. 

I don't mind a bit of poetic license but I'm waiting for the 90% cost saving scenario  in detail. And let's not forget that Scott is not an impartial obserever here.

And just to note, I have no reason to defnend the ongoing use of routers -- the sooner that SDN and NFV can be implemented in a secure fashion, the better for the operators. Bring it on. But the timescales and extent of the mix of virtual/legacy is not going to suit the aspirations of the vitrualization flag-bearers any time soon, I don't think. 
jabailo 11/18/2015 | 2:50:32 PM
DSLAMs ? What about the D-SLAMs that are 100 ft to 300 ft from every home where fast broadband is available?  Aren't those routers?  How can they be changed from hardware to software?  You'd still at the very least need a swtich to connect twisted pair to fiber.

And when there's fiber to the home, do you still need a local router?  Or does the signal just travel much further upstream to the CO?

SeniorMa50381 11/18/2015 | 12:54:18 PM
RIP Router I remember many years ago at the dawn of the switching era - a certain company touting the death of the router, Where are they now? However I agree that routing has been overused at the edge for public (internet) routing but it is useful to separate the SP routing domain from that of the customer.
Page 1 / 2   >   >>
Sign In