Is MPLS dead? again?
In the age-old battle of netheads vs. bellheads (look at the date on this wired article), I've always been a nethead. This might seem strange for someone who's worked for telcos for most of his life, but I've been mostly on the IT side of the business and saw the network as something basically dumb that just ought to get on with its job. The recent Gartner article (warning, behind Gartner paywall) seemed to me to be starting from the nethead perspective but swings rapidly into bellhead territory by making the case for intelligence within the network infrastructure.
This is not a new debate. It got me thinking about some of the other classic internet memes and how they are coming together to point the way to how we will build secure infrastructure over the next development cycle.
the cathedral and the bazaar
Originally about the way Linux was developed, the idea that a loose collection of individuals can, with the correct motivations, form an open development group that develops and perfects something has become enshrined in a whole host of others. I consider that Agile is more a product of bazaar thinking than anything that the clergy in a cathedral would have come up with.
software defined networks
This is about making the underlying fabric of the network dumb, extracting all the intelligence into the network operating system. Sounds like a stealth attack by netheads. In the future, it means that the same fabric will be able to provide internet and MPLS-type services.
mobile first design
A very basic design concept, originally for the Web but now being applied to most application development, where the developer starts with the requirement to deliver an application to a simple, mobile device, then adds functionality and design elements to take advantage of richer clients.
Clarke's Third Law: "Any sufficiently advanced technology is indistinguishable from magic" - Arthur C Clarke
If the network is doing its job properly, that is, delivering applications with the appropriate quality of service to give the best end-user experience or efficient machine to machine communications, then it will become invisible. "It just works" is what we are aiming for here. One of the best mission statements I ever heard was from the IT operations director of a water utility in the UK who said "When our customers turn on the tap, water comes out". That's the experience that our customers want.
so what does a magical, invisible network look like?
Firstly, the nethead view is in the ascendant. The objective is not to build the best network. What our customers want to do is to deliver applications so how the network technically does its job is becoming less important.
Our view of the future of networking for most enterprises is
- a hybrid of the public internet (with CDN and optimisation technology to make the experience as good as it can be)
- carrier-grade internet (with carefully selected peering points for access to low-latency huge bandwidth)
- and MPLS (for workloads that have to have end to end quality of service).
All of this networking will have the appropriate security controls, SLAs and integrated with public and private Cloud infrastructure as part of the deal and will be able to deliver to mobile and fixed end points.
what does this mean for communications service providers and their customers?
Competing by having the best coverage and most reliable MPLS network will not be enough. CSPs will have to gain competitive advantage by integrating all the components required for the hybrid world. The way we bring everything together will be much more a bazaar than a cathedral as the ability to centrally plan, specify and build a single infrastructure becomes less important than the ability to bring all the components from various third parties, in all geographies, together to meet each individual customer's requirements. Here's a view on hybrid from Ipanema, one of our technology partners.
At Orange we have invested a lot in partnerships with Akamai, Microsoft and many others to put together our secure application delivery infrastructure approach for our enterprise customers. More about this in the coming months.
For our customers, it's time to put your thinking caps on. As hybrid takes hold it will become increasingly more complex to design, build, implement, manage, and optimise an IT and communications infrastructure. Some of you will see advantage in operating the infrastructure or you may have regulatory constraints, but we think that many of you will give the pain to a service provider like Orange.
so, is MPLS dead?
Simple answer is no, it's just becoming an invisible part of what delivers "water to the tap".
crédit photo : © NLshop - Fotolia.com
July 19, 2013
The good news at the mobile end is that 4G/LTE and the coming 5G have much better data network performance, especially in terms of latency, than 3G so that the end user experience should be much better for mobile application delivery. I believe that MPLS/tagging will continue to be important in the carrier core and Cloud services provider networks (which might be the same organisation) as networks converge onto aggregated SDN backbones. The thing to watch is the emergence of standards for interoperation of data centre and carrier SDNs. If we can extend tagging to the mobile delivery part as well, that can only be a good thing.
July 17, 2013DrrobevansI agree that networking will melt into the background and MPLS will just be the technology that makes things work in the background. Networking is already seen as just another utility like electricity, water or gas, it just works and it's invisible, yet visible when it's not there (oh the irony). Ethernet will end up being the technology of choice, but MPLS will run over Ethernet (EoMPLS). Also agree there's pain, but the security concerns of enterprises means that a lot of control won't be handed over. Ultimately, as we shift to a mobile world the questions becomes, will we still need MPLS/Ethernet in the enterprise for workforce connectivity?