So I know what the MPLS technology does. But say a business has an MPLS circuit. I work in a NOC where the majority of clients are SDWAN clients using a vmware veocloud. We always have to ask the customers to perform layer 1.
So where does an MPLS circuit plug into, just a regular ATT modem and to the velocloud? Or is there something I’m not understanding. I’ve heard a MPLS circuit is virtual? Can someone clear up this concept for me on how the physical connections are.
The service provider uses MPLS forwarding in its network and most likely it’s for nearly all traffic.
The sales pitch of SDWAN to replace MPLS circuits is to replace specifically engineered virtual paths through the network that are sold to customers. This will normally have a number of predefined and sla backed paths and traffic characteristics that make these high value circuits and thus they are sold as such.
If you don’t buy an MPLS circuit and just use an internet circuit it’s highly likely that you do traverse an MPLS network but get none of the traffic engineered goodness.
Thus sdwan leverages multiple providers and link quality monitoring amongst other things to try to emulate the MPLS quality over multiple best effort networks. And for the vast majority of use cases it does a great job.
When customers talk about “MPLS” is most often actually a Layer 3 VPN (AVPN in AT&T’s case), there may or may not be label switching going on inside the SP but that’s not important and isn’t the service the customer is paying for.
In all cases there will be a CPE at the customer site doing the (typically Ethernet) handoff to the customer. What happens after that is up to the customer.
I think ADI is “ATT Direct Internet”. Not sure about ATI. I think AVPN/MPLS has been covered. Deciphering an ATT invoice is never fun.
In my past experiences MPLS circuits I have used some sort of router supplied by the vendor and that circuit then either plugged into a firewall our router on site that belongs to us. When I did a full implementation of SDWAN we just moved that circuit from our firewall to the SDWAN device and made the proper routing changes and everything works fine. It’s not too complex once you understand how things are connected
MPLS mostly runs over Ethernet these days, like everything else.
Layer 1 is just 1’s and 0’s, not sure what you mean when you say that’s what you ask the customers for. Networking requires framing / data link at least.
But yeah, a virtual L2 or L3 service, delivered by a carrier leveraging MPLS, will typically be delivered over fiber using one of the optical Ethernet PHY standards like 10GBase-LR. L2 is L2, and L3 will normally run BGP between customer site and carrier router.
MPLS is way more reliable than standard Internet connections (you still need your Edge Router → Internet connections, but for Satellite Offices, MPLS can be a more reliable option). Internet connections can be unstable sometimes, and most of all, you have no control over what happens beyond your Edge Router. Your outbound traffic can bounce between carriers and several ISP Routers before reaching the destination
A Large benefit of MPLS is that you’re paying for a dedicated link from 1 ISP, and you can at least know where that traffic will traverse beyond your Edge Router. Saves you from the trouble of a far upstream issue that you have no control over, that your users are affected by
In terms of how the physical connections work, you have your MPLS Router(s) that connect to one of the MPLS provider’s Label Edge Router (LER), which services multiple clients. When you send traffic to their LER router, it places a label that determines the path to the destination. Once determined, it sends your traffic down a path to a destination, through a series of Routers known as a Labeled Switch Path (LSP)
As opposed to Internet Routing, where all traffic, from your Edge Router to the destination, is based off of L3 Routing protocols
*Edit : Grammar
An MPLS service is going to have the same kind of handoff/physical connection as typical dedicated Internet access. Most likely this is an Ethernet handoff. It may be electrical (CAT6) or optical (single mode or multimode fiber using pluggable transceivers). Physically, you can expect the same kind of connection that you’re used to.
We have Velocloud that has MPLS and Internet circuit. MPLS hand-off is fiber but we also have copper at legacy WAN sites. ADI is AT&T dedicated Internet which is AT&T dedicated Internet product.
Excellent description, subtle inclusion of the best effort traffic is likely in its own overlay network.
Love this. Beautifully said 
Edit: Marketing lies have me fooled
Layer 1 refers to the physical media. Copper, fiber etc… The 1s and 0s would belong in layer 2 data-link. That’s where the negotiation of how data is exchanged and encoded occurs.
So check layer 1 can be directly translated to, is it plugged in…
Agreed. Also there’s cost efficiencies/benefits with internet circuits vs MPLS.
No, MPLS is literally label switching instead of packet switching. There are many technologies that people incorrectly stick under the “MPLS” name. psudowire, EVPN, vpls, etc.
Exactly, in NOC’s saying “customer to perform layer 1 troubleshooting” is code for the customer is going to verify power to equipment, nothings unplugged, and most likely a reboot of equipment
That’s not correct.
Layer 1 is mostly about how you represent 1s and 0s over a medium. So things like voltage, light levels, and line coding schemes like differential Manchester, NRZ, PAM4 and all those things. In our era the ETH PHY basically.
Data link layer is about framing and structure, how two devices make sense of the raw bitstream layer 1 provides.
This statement is somewhat misleading. VPLS at least in Nokia and juniper terminology reference an MPLS service topology same with puesdowire and other vendors. The EVPN address family can support MPLS tunnels or VXLAN.
Sure maybe people say that. But you shouldn’t be confused about what layer-1 means in data communications. It very much involves how bits are encoded and transmitted over the medium.