Several T1 links to one Ethernet hand-off

mzinz

Is this a common scenario in the real world?

A customer is switching to an MPLS topology. There are several remote sites, lets say 20. Each remote site has been provisioned to connect to the hub router via a point-to-point T1 link.

I found out last night that they won't be terminating back to a channelized T3, though - there is actually a 50M ethernet handoff.

Is this a normal setup? I wonder what the benefits of this are over a T3?
How does the ISP handle this? Do they have a large router that is actually using a T3, then that router connects straight to our hub router?

Any info or links would be helpful

networker050184

Its a common set up that I have seen and most everything new getting truned up besides a T-1 is going the way of ethernet.

mzinz

networker050184 wrote: »

Its a common set up that I have seen and most everything new getting truned up besides a T-1 is going the way of ethernet.

The thing that surprised me was that all the links are Point-to-point T1s, so I'm wondering how they are all converging.

On my Ethernet handoff will I just be able to sub-interface it and have individual point-to-point IP's on each subinterface which point at the T1 endpoints?

networker050184

mzinz wrote: »

The thing that surprised me was that all the links are Point-to-point T1s, so I'm wondering how they are all converging.

On my Ethernet handoff will I just be able to sub-interface it and have individual point-to-point IP's on each subinterface which point at the T1 endpoints?

Ah, I get what your question is now. With the MPLS technology the T-1 sites will just terminate on the service providers router, not actually terminate a point-to-point circuit from the remote site to the hub. The service provider will then carry your reach ability information across the MPLS cloud for you. Every site can have a different L2 access method without issue.

mzinz

networker050184 wrote: »

Ah, I get what your question is now. With the MPLS technology the T-1 sites will just terminate on the service providers router, not actually terminate a point-to-point circuit from the remote site to the hub. The service provider will then carry your reach ability information across the MPLS cloud for you. Every site can have a different L2 access method without issue.

Okay, so yeah that was exactly what I was wondering

When you say that the T1 will terminate on the ISP router, are you referring to both ends of the T1's, or just the hub end? Ie: Will the remote ends still do T1 termination?

I'm assuming that the answer to my other question is yes, but, since the aggregated T1s will be handed off via Ethernet, will I be able to just sub-int the ethernet interface and assign a different point-to-point IP to each one?

networker050184

You will have to speak with your ISP, but it sounds like you will not be terminating the sites to your main site. It will be a point-to-point link from each site into the "cloud" in which you will not handle the terminating end. It will be something like:

remote sit T1===PE router===CLOUD===PE router===Ethernet Main site

Each sites link will be a point-to-point to the service providers network NOT to your other site. The service provider will carry your information inside a VRF and **** it out to each site as needed. This is one of the benefits of MPLS VPNs that you don't have to buy expensive point-to-point links between your sites.

Again, this is what I'm assuming from my experience and your description. You should talk to your provider for details and confirmation.

mzinz

networker050184 wrote: »

You will have to speak with your ISP, but it sounds like you will not be terminating the sites to your main site. It will be a point-to-point link from each site into the "cloud" in which you will not handle the terminating end. It will be something like:

remote sit T1===PE router===CLOUD===PE router===Ethernet Main site

Each sites link will be a point-to-point to the service providers network NOT to your other site. The service provider will carry your information inside a VRF and **** it out to each site as needed. This is one of the benefits of MPLS VPNs that you don't have to buy expensive point-to-point links between your sites.

Again, this is what I'm assuming from my experience and your description. You should talk to your provider for details and confirmation.

Thanks for all the info - that does make sense.

One last question I had, and I may need to talk to the ISP about this: Are MPLS implementations like this usually end-to-end? Ie: should I be configuring MPLS on my remote routers and hub router, or is it something that only the ISP will implement in their backbone, or is it up to me?

networker050184

mzinz wrote: »

Thanks for all the info - that does make sense.

One last question I had, and I may need to talk to the ISP about this: Are MPLS implementations like this usually end-to-end? Ie: should I be configuring MPLS on my remote routers and hub router, or is it something that only the ISP will implement in their backbone, or is it up to me?

You will not configure anything concerning MPLS. The ISP will take care of all that. It will look just like any other WAN connection on your end.

mzinz

networker050184 wrote: »

You will not configure anything concerning MPLS. The ISP will take care of all that. It will look just like any other WAN connection on your end.

Kinda dissapointing I was hoping to play with MPLS.

When would a customer get to configure it? If they had supplied nothing but point-to-point T1 links?

dtlokee

In most MPLS deployments the customer will only need to configure the circuit parameters (linecode and framing on a T1 for example), the encapsulation (MPPP in you NxT1 deployements) and BGP peering if it is provisioned this way some are static routing in the carrier network but most I have seen are BGP. MPLS to you will simply look like a L3 transport between your sites, that is why you can use a T1 at one end and an Ethernet circuit at the other, the transport is L3 not L2 and is independent of the L2 circuits in use.

When would an enterprise network build an MPLS network of their own? Almost never, there are some rare occurances where it is desirable to do so but the main advantages of MPLS are going to be realized by a service provider who wants to transport IP packets from many customers that may have overlapping IP address ranges over a shared backbone. This is the strength of MPLS and doesn't really describe most enterprise networks, even large ones.

mzinz

dtlokee wrote: »

In most MPLS deployments the customer will only need to configure the circuit parameters (linecode and framing on a T1 for example), the encapsulation (MPPP in you NxT1 deployements) and BGP peering if it is provisioned this way some are static routing in the carrier network but most I have seen are BGP. MPLS to you will simply look like a L3 transport between your sites, that is why you can use a T1 at one end and an Ethernet circuit at the other, the transport is L3 not L2 and is independent of the L2 circuits in use.

When would an enterprise network build an MPLS network of their own? Almost never, there are some rare occurances where it is desirable to do so but the main advantages of MPLS are going to be realized by a service provider who wants to transport IP packets from many customers that may have overlapping IP address ranges over a shared backbone. This is the strength of MPLS and doesn't really describe most enterprise networks, even large ones.

Excellent response - thank you for the detail.

mzinz

I assume that on the ethernet handoff end, I will just have multiple subinterfaces configured with the Point-to-point IPs?

ie: fa0/1.1, ip address x.x.x.x y.y.y.y
fa0/1.2, ip address z.z.z.z y.y.y.y
etc
etc

networker050184

On the ethernet hand off you should have just one ip like a point to point circuit. The ISP will handle routing between the site. You are still thinking in terms of leased lines, which is not what MPLS is used for.

mzinz

networker050184 wrote: »

On the ethernet hand off you should have just one ip like a point to point circuit. The ISP will handle routing between the site. You are still thinking in terms of leased lines, which is not what MPLS is used for.

Ok - I see what you are saying now.

And yes, I can't get leased lines out of my head

You're saying that on my hand-off end, I configure a single IP like a regular L3 link. I should be able to get to a remote (private) subnet just by creating a static route, then, correct? (to the remote public T1 IP, using L3 hand-off as gateway)

SysAdmin4066

That's correct, the routing for all of the subnets is performed by the provider.

e24ohm

Are MPLS, similar to EFM links I have been reading about?

Thank you.

networker050184

EFM is just Ethernet in the First Mile (or last mile as most providers would refer to it). You could have an MPLS VPN solution handed off to you on ethernet, but they are not the same thing.

e24ohm

networker050184 wrote: »

EFM is just Ethernet in the First Mile (or last mile as most providers would refer to it). You could have an MPLS VPN solution handed off to you on ethernet, but they are not the same thing.

understand...thanks.

WRKNonCCNP

mzinz wrote: »

Ok - I see what you are saying now.

And yes, I can't get leased lines out of my head

You're saying that on my hand-off end, I configure a single IP like a regular L3 link. I should be able to get to a remote (private) subnet just by creating a static route, then, correct? (to the remote public T1 IP, using L3 hand-off as gateway)

Note that you will have to advertise the subnets on the customer side of the routers terminating the T1s and the 50 M Ethernet to the provider MPLS network via [probably] BGP. In order to pass traffic between the customer subnets, they must be advertised into the provider's backbone routing protocol.

You cannot simply create a static route to the provider-side IP of the remote site router in order to reach a customer subnet behind that remote site router. The IP addressing on the interfaces terminating the T1s and the Ethernet won't be in the same subnet. I may have misunderstood you previous comment, so sorry if i just told you something you already know. Also, you will probably be using a single subinterface on the router terminating the 50 M Ethernet since you will probably be doing 802.1q trunking as well.

kryolla

Since you are using leased lines on the remote sites and an ethernet handoff on the hub side the provider has be to aggregating all the leased lines and converting it over to ethernet, probably a card in the mux is doing this. See if you can take a tour of the provider co-locate facility and will see either fiber or co-ax coming in and ethernet coming out. The rest is basic MPLS with a CE to PE routing protocol and BGP VPNv4 between the PE routers with MPLS as the transport.

networker050184

kryolla wrote: »

Since you are using leased lines on the remote sites and an ethernet handoff on the hub side the provider has be to aggregating all the leased lines and converting it over to ethernet, probably a card in the mux is doing this. See if you can take a tour of the provider co-locate facility and will see either fiber or co-ax coming in and ethernet coming out. The rest is basic MPLS with a CE to PE routing protocol and BGP VPNv4 between the PE routers with MPLS as the transport.

From the sounds of it I don't think this is whats going on here (I may be wrong of course). It sounds like the OP is getting a standard MPLS VPN set up which will not require ethernet to TDM conversion. If it was an old school VPN type set up then he would need the conversion to terminate the T1s at the hub, but since its MPLS VPN the T1s will terminate on the local PE and the rest will be carried over IP.

kryolla

Sorry I shouldn't have used conversion as we are talking about layer 1 and 2. if there is leased line on one side and ethernet on the other side, somewhere in the provider network the leased line has to be aggregated and ethernet hand off. Which Im guessing will be cheaper than getting a DS3 on the hub site. Im curious to know if all metro-ethernet hand offs are in a telco building or business park with all the major telco or close by it as they would use them for layer 1 transport. If they are not then they would have to order leased lines from the local telco.

networker050184

kryolla wrote: »

if there is leased line on one side and ethernet on the other side somewhere in the provider network has to be aggregated and converted. All this is transparent to the customer but all happens on the provider side

There is no need for conversion. The T1 will terminate on a router and the ethernet hand off will be terminated on a router. All traffic will be routed over the MPLS backbone. I'm actually in the process of integrating a new device that will do exactly what you are talking about, but is not for this purpose, its not needed in this situation. We are going to use it to aggregate T1 customers onto an ethernet port on a router which costs a lot less than a T3 card form Cisco.

kryolla

I edited my last post I shouldnt have used conversion

The T1 will terminate on a router and the ethernet hand off will be terminated on a router

what about in between, you have mpls in between but you are not running T1 all the way and a 50meg hand off so you are aggregating at the PE with multiple CE or at the P with multiple PE circuits. So it might be T1 at the edge but at the core it might be DS3 or optical OC3 or higher

That should be a nice project but its not a Cisco piece of equipment . Good luck sounds like a fun project. The company I work is no longer in a build phase so there is a lot of downtime which I use to study for IE but I sure do miss those projects