Insight

MPLS vs. E-LAN (VPLS)

Terminology seems to be the complete bugbear of our industry at present. Cloud this and virtual that, it’s really beginning to become white noise, a sort of theme tune to the Ethernet market. According to Sachin Vaish, Managing Director at Vaioni, it seems more and more that those who succeed are those who are capable of cutting through the lingo to the heart of the problem to deliver a solution to a client’s needs.

Last year Sachin Vaish produced a table to assist resellers in choosing the right kind of connectivity for their users. Today Sachin has a new table, this time looking at WAN solutions.

Vaish says that as far as WAN solutions are concerned there really are two main contenders, MPLS (Multi-protocol Label Switching) and E-LAN (Otherwise known as VPLS - Virtual Private LAN Service).

“These are two very different solutions, sometimes using different equipment and technologies but often achieving similar goals.

At Vaioni we’ve planted our flag firmly in the VPLS camp. Does this mean we can’t do a MPLS like solution? Does this mean that VPLS is inherently better? No. We’ve been driven by modern customer demands and therefore built an infrastructure that delivers on both price point and SLA, thus allowing us to complete with the larger operators. For us MEF (Metro Ethernet Forum) Compliant E-LAN in conjunction with VPLS was the best tool for the job. So let’s look at the main differences.

 

Feature Comparison

For the purpose of this piece I’m comparing Layer 2 MPLS with VPLS based E-LAN. Layer 2 over MPLS technology such as AToM (Any Transport over MPLS) has been ignored.

MPLS

MPLS has been around a while now and for good reason. The technology is extremely mature and reliable and well supported across most vendors. It also fits with the needs of most customers. That said MPLS is showing its age. It’s use in the control plane in the form of LDP/LSP (the MPLS path itself) is here for the foreseeable future with emerging technologies such as OpenFlow far from production standard or if it will ever become the standard.

MPLs will in turn give the network operator excellent control of their data plane, with resilient paths and Layer 3 Meshes that can be deployed both quickly and safely, with the onus of inter-site routing put largely on the shoulders of the provider.

 

E-LAN (VPLS)

E-LAN isn’t so far removed from its parent MPLS. In fact VPLS is dependent upon label switching for operation. As such VPLS can be seen as an extension to the MPLS toolset. The more general term of E-LAN is not itself as dependent upon MPLS with other technologies such as MEF compliant switches able to perform the tasks without MPLS.

On top of the normal routed infrastructure as offered by a traditional MPLS, E-LAN allows the control of the routed infrastructure to be dictated by the end user. Optionally some providers can provide a level of Layer 3 (IP address) management for those users not comfortable doing so themselves.

Why should I care?

I assume some of you are asking yourself this very question right now, and it’s a very good question. Largely the technology used by the network operator should be your concern. The network operator should act as an agnostic layer of intelligence, responsible for ensuring that packets get to where they need to be, and only where they need to be. The means by which this is done can only affect you in two major ways and comes down to the old adage of Cost vs. Quality. This is where the main forces within the market differentiate themselves.

There are those network operators in the market place who have literally spent millions upon their infrastructure only to receive something of little commercial value but extremely reliable. Then again I’m sure we can all name some very large outfits who’s infrastructure seems to cause an unending stream of faults and service outages, with seemingly small issues having large impact craters upon customer service levels.

Speaking on behalf of Vaioni as a smaller network operator, we’ve gone for the niche of getting a lot of functionality from a sensible amount of investment. This has allowed us to easily integrate things such as UTM (Unified Threat Management) security, VoIP and Storage within the WAN product portfolio without costs getting out of control, but whilst maintaining reliability, guaranteed Quality of Service and minimal hops across the applications.

When choosing your network operator, look carefully at the offering. If they only offer MPLS, chances are you’re sat on an old infrastructure incapable of VPLS that’s being milked for the return on investment. You’ll probably find the service reliable, maybe a little slow/contended at times, but you shouldn’t be hit by incessant maintenance windows whilst bleeding edge features are patched. Of course ringing up said supplier and asking for new features will result in excuses. Any features possible from the system would’ve been released long ago if they were available.

Of course on the flipside if your network operator sounds like a yes man, be prepared for disappointment. They may be selling features that simply aren’t production quality. At which point you’d better hope they have good support from their vendor, as you’ll be in maintenance hell as they push out patch after patch to fix yours and other customer issues.

The Hybrid Approach

This takes me back to last summer when my nine year old decided to build a time machine from my old golf bag and bits of wood lying around the workshop. As I came in I found him hammering a screw into the golf bag with a metal work hammer. It was time for my son to learn about ‘using the right tool for the right job’, and time for me to buy a new golf bag.

We can all laugh at the above example, but there are operators and ISP’s we are all using today doing that exact thing. They keep hammering away at that old kit and network, crowbarring it into doing something that it was never designed to do. When operators and ISP’s built their infrastructure, Layer 2 private circuits weren’t even a thing. You delivered physical circuits, or you gave the customer some IPs and away they went.

 

 

To put it another way, who else talks about E-LAN technology? There’s not many of us out there, but I’d guess the name on all of your lips is BT. BT wear this product name with pride, but are expensive. E-LAN isn’t a product name, it’s a standard. The standard is set by MEF and it means that a certain set of standards is being adhered to. The customer can be assured that their products will work in the correct manner. BT spent a lot of money on the 21CN network to get these products, they could’ve kept with the old platform, but they like us saw how wholesale Ethernet was changing and moved quickly to adapt.

So at Vaioni we stand by this standard too, that doesn’t mean we won’t sell you some MPLS. If you want ADSL or 3G services integrating into your mesh, MPLS is the way to go, but even these old Layer 3 technologies are changing. For example take GEA (General Ethernet Access), look closely and you’ll find it’s simply Layer 2 FTTC.

As such there’s no problem with MPLS per se, and in a lot of circumstances we’d happily deploy it, but we’d much prefer to deploy you a VPLS with routing on top. This means that when the day comes the customer decides that the SAN subnet needs to span two sites so replication is better; you aren’t re-designing the whole solution. You simply turn up another VLAN at each site. They keep their original routed VLAN and boom. Site to site SAN replication with Jumbo frames with the sub net spanning both locations. The time for rollout is reduced to 48 hours as opposed to weeks to convert a Layer 3 MPLS to VPLS.

This hybrid approach, mixing routing MPLS and E-LAN allows for your customers network to rapidly change shape according to their needs. This in turn allows for the client to get creative with their virtualisation needs and newer technologies in that arena. One such example we’ve seen a lot recently is that by spanning a subnet between a customer’s HQ and data centre based co-location space, customers can teleport Virtual Machines quickly to a central location in case of HQ failure.

Conclusion

So what should you take away from this piece? Well I hope more than anything you’d draw your own conclusions. Don’t trust those who promise the world, and those flogging a dead horse will cause you problems later on. The sweet spot as always is the middle ground. What we’ve attempted to do is to make all the carriers behave in the same way, smoothing off the rough edges of the carriers with older networks and really pushing the limits of those carriers with newer and better infrastructures. From this we’ve given ourselves best of breed connectivity and the pricing. It takes some good engineering and knowledgeable sales staff that understand but don’t over sell.