Maintaining MPLS VPN service consistency has emerged as a major issue for companies employing multiple carriers for global reach. We talked to Charlie Muirhead, CEO and founder of Nexagent, a five year old UK company that develops hardware and software designed to link carrier networks.
Is MPLS VPN service consistency a big concern among corporations?
The challenge is connecting networks in a scaleable and reliable way while ensuring that you can extend the rich service model from one network to another, rather than diluting it down to the lowest common denominator services. How do you solve this from an operational and process standpoint? There's a whole bunch of choices available to them. Historically, enterprises pretty much built the solutions themselves. They would buy bandwidth from many carriers. That was a big task.
Along comes the world of VPNs and the large international carriers would turn around and say, 'If you want consistency and service-level agreements, you have to give it all to me, on-net.' That jarred some customers who played their bandwidth suppliers off of each other to keep them keen on price and quality. So these global enterprises find themselves in an interesting position: Do they go to a single global carrier knowing that that carrier doesn't have the network in all the countries, and somewhere under the hood there's an activity going on which is really network integration?
The default is to continue doing it yourself. That means buying MPLS services from different carriers around the world and integrating those. The third option is to go straight to a systems integrator who historically would essentially take over running your in-house do-it-yourself architecture with great scale of economy. Then you have the virtual network operators that are just beginning to emerge.
Sounds like a daunting problem for a company.
It is. The global enterprises have a lot of choices. There's pros and cons to each. They're always quite suspicious about the motivations of the global carriers because they're always trying to get your traffic onto their facilities so they get their margins. And they're always a bit suspicious about what's actually going on under the hood and where do they have network and where don't they. Most of the time you'll hear about carriers responding to bids with pricing for geographies they say they're going to have when it's not quite in place yet.
How do interconnected carriers hash out the billing and settlement issues between them?
Let's say a global carrier wants to have partners in five off-net regions. Their requirements will be to link up carriers to their own core network in a secure and reliable way, where they can put multiple customer connections to the same physical interconnect. So they need to have a method of physically connecting those and controlling which VPNs extend across which partnerships. That requires quite a bit of network architecture to be done. Then you need a way to run services that go across [those interconnect points]. [Carriers] need to do a pre-sales process of examining which partners they're going to have to use in which countries and they have to come up with some way of modelling that.
Those pre-sales processes have got to be pretty thorough because you've got to come up with a price for the solution that you can bid to the customers to make you money, but still keen enough to win the business. You've got to have the tools to calculate what needs to be done on your partners' networks. The instruction you send to your partner is typically a work order, which says 'I want this service from you in these countries, on these parts of your networks, with these SLAs for these sites and these three addresses, and this kind of monitoring frequency and so forth, and here's the money we're going to pay you...'
That's the kind of level of detail that would go to each of your carrier partners and then that would all need to go into a project plan. And you need to track the progress of each of your carrier partners around the world to make sure they deliver what they're supposed to deliver on time. So there's a whole assurance process that goes on which feeds up into the macro-assurance process that the global carrier will already have in place.
You recently announced that you joined the MFA Forum, which is also working on an MPLS Interconnect specification. Why?
Four years ago, we delivered Version 1 of our interconnect architecture. Over the course of those years, it's evolved. Now that the industry is waking up to the challenge, we wanted to get the appropriate forums thinking about the problem and agreeing how that should look in the future. We absolutely support an open interconnect architecture and our software over time has evolved to support whatever comes through those standards processes.
We are participating and contributing as much as we can of what we've learned into that process so that we get the best brains in the world thinking about what this should look like so we can come up with something for the industry. That, of course, will be an interconnect point that enables the carriers that have agreed on standardised service definitions to work very easily together. We're also jointly proposing a new working group in the IETF with a handful of vendors and a handful of the carriers on this area of VPN service interconnection.
You're also a member of the IPsphere Forum, which is also addressing the interconnect issue. Are these activities redundant?
They're all doing different things. The IETF is working at a protocol level. The MFA is about implementation, best practices at the network layer - what they call in the IPsphere the ICI, Inter-Carrier Interconnect. The IPsphere is working at the services architecture level saying, 'What services do customers want to buy? What will those look like?' and how do the different partners signal to each other what they want. So they all have different functions and this is such a huge space that there's no one organisation that could tackle all of that.