Service Providers

pmoyer

Is MPLS Dead?

by pmoyer ‎12-17-2013 11:39 AM - edited ‎12-19-2013 10:24 AM (2,759 Views)

Last month marked the “16th Annual International Conference on Next Generation Internet and Related Technologies”; which is hosted annually in Washington, DC. That’s a lot of words to describe the annual Isocore sponsored MPLS conference. For those that have not followed this annual meeting of the MPLS minds, this has traditionally been a full-on, MPLS-centric, love-fest of sorts. Although last years conference did not explicitly have “SDN” in the title, it was held in conjunction with an “SDN & Inter-Cloud Summit”. At this year’s event, there is no hiding the fact that this traditional MPLS-centric conference has now become an SDN + MPLS conference. Has SDN finally arrived?

 

Here is the link to this year’s conference.  If you take a look at the technical session agenda, it becomes quite obvious that SDN is a primary focus of this conference; although, MPLS is still heavily represented. So, what’s my point?

 

Well, I think we have experienced an inflection point of sorts. By “we”, I’m referring to the networking industry in general but more specifically, the networking industry that has supported all the MPLS technologies and solutions over the last decade and a half. The speaker and attendee crowd at this annual Isocore event has been driving the MPLS agenda over the last 15+ years and has (finally?) turned the corner to embrace and address the SDN evolution that sits before us. There may not have been an outright admission of an SDN inflection point, but the sense from the presentations, the Q&A and the hallway chatter is the realization that SDN is here, SDN is our future, and that the industry has been overly focused on MPLS-related technologies for too long. This MPLS-centric view of the networking world has been too narrowly focused and the crowd that has been driving the MPLS evolution (from Frame Relay and ATM networks to MPLS) has been somewhat naively ignoring some of the serious and glaring architecture problems that are present in the networking industry. And mind you, I am a huge fan of MPLS so I am not bashing MPLS by any means!

 

To add some additional color to this realization, one of my colleagues recently jested during a talk he was giving on SDN that “we, in the networking solution space, should be somewhat embarrassed by what’s happening around us”. He was specifically talking about the Open Networking Foundation (ONF), but I think his comment applies equally across the networking industry.  His point is that the ONF was founded by service providers and came into being because the currently available solutions are not adequately solving some of the enormous challenges and problems they are facing. To relate this back to the point I am making; the “MPLS hammer” has been overused for too long, and every new problem has been looking like just another nail.

 

While not entirely new, two examples of this evolution can be seen with the IETF I2RS WG efforts and the emerging DC multi-tenancy overlay solutions. The I2RS activity is addressing the need for additional programmatic control of the network; in other words, allowing network applications to dynamically augment the routing control plane. The emerging DC multi-tenancy solutions provide logical overlay networks to meet the multi-tenancy requirements in cloud data centers. One focused industry effort to address this DC multi-tenancy requirement is in the IETF NVO3 WG.

 

So, where do we go from here? I believe we are heading in the right direction and this belief was confirmed at the recent Isocore event. The atmosphere and growing consensus among some of the smartest people in this industry is the recognition that networking finally needs to evolve beyond MPLS. This is not the same as the industry consensus of 10-15 years ago, when networking clearly needed to evolve from ATM to MPLS. SDN is not going to replace MPLS; that’s not what I’m saying. MPLS continues to solve real problems. But the industry is evolving towards more software-defined and software-driven technologies and solutions.

 

So no, MPLS is clearly not dead. But SDN has undoubtedly arrived …