Moving from T1 (point-to-point) to Metro E

Q

Quikstrumental

Limp Gawd
Joined
Oct 18, 2005
Messages
336
Hey guys,

I'm attempting to move our network from our existing point-to-point T1s, over to Metro Ethernet.
From what I was told, I could have all of our sites' routers use an IP address with the same subnet mask, and they should all be able to communicate. These are Cisco 2800 series routers.

Example:
Router 1 (10.0.0.4 /25) connected to Metro E at Site 1.
Router 2 (10.0.0.5 /25) connected to Metro E at Site 2.

I am then able to ping each router back and forth from both sites.
This is where it gets tricky: I go to Site 3 and add Router 3 with an IP address of 10.0.0.6 /25, now all of the devices can ping each other, but then they all start dropping packets and everything becomes unstable.

I am currently using RIPv2, but I have heard of other companies using OSPF or EIGRP. I have not tried any other protocol, but I would like to know if I am headed in the right or wrong direction.

I could draw up a quick diagram if needed.
 
What type of MetroE product did you purchase? Is this a hub and spoke product? Any to Any, or point to point? Most of my customers front their MetroE service with a router due to mac limitation within the core. Having said that, you do not need a router at the edge for MetroE to work. MetroE is a flat layer 2 service so if you, in theory, plug in a laptop on either end of the circuit, IP each laptop within the same subnet, no gateway, you will be able to ping back and forth.

As far as dropping packets? What are you using to determine packet loss? What size is your circuit at each location? Ripv2 uses multicast to send its updates and these updates are also part of RIPs way to send "keepalives." Every 30 seconds at that. Doubtful but you may be saturating you link by using RIP. I personally would use OSPF, or if you need the extra fast convergence time of EIGRP, cisco propriretary so you will need a cisco router on each end, then go with that. A bit more info is needed.
 
Some(most?) Metro circuit providers do not allow multicast across their WAN. Thus your RIP updates might be having problems.
Since everything is Cisco why are you not running EIGRP. It's so simple to setup.
 
Hey guys, thanks for the quick replies

Mikey, I believe they said any-to-any MPLS. I'll have to verify with my provider. I was simply pinging between sites, and the response times would increase exponentially, and then time out. Then start once again. I'm attempting to keep the routers at the sites since they're on their own LAN subnet.

cyr0n, I've barely looked into OSPF or EIGRP today. I was hoping the [H] would confirm or deny the attempt to try OSPF or EIGRP. I mean, I can ping the routers I add on, but they just make the system trip out and lose connectivity when there's more than two.
 
cyr0n_k0r said:
Some(most?) Metro circuit providers do not allow multicast across their WAN. Thus your RIP updates might be having problems.
Since everything is Cisco why are you not running EIGRP. It's so simple to setup.
Click to expand...

Agreed, and I should have known this as we block multicast traffic. Broadcast traffic is limited.
 
Last edited:
Quikstrumental said:
Hey guys, thanks for the quick replies

Mikey, I believe they said any-to-any MPLS. I'll have to verify with my provider. I was simply pinging between sites, and the response times would increase exponentially, and then time out. Then start once again. I'm attempting to keep the routers at the sites since they're on their own LAN subnet.

cyr0n, I've barely looked into OSPF or EIGRP today. I was hoping the [H] would confirm or deny the attempt to try OSPF or EIGRP. I mean, I can ping the routers I add on, but they just make the system trip out and lose connectivity when there's more than two.
Click to expand...

Ok cool, we call any to any, ENS or E-Lan as named by the MEF. You DEF want to front each site with a routers as with any to any, there is typically a MAC limit at the PE router. I know for us, its 250 mac address per UNI and 2500 per ENS domain. In, say, a hospital environment, you will blow past that mac limit and start to notice machines can't talk across the network as their macs will not register until another machine times out or shuts down.

Also, have the provider check their CPE device that attaches to your router. More than likely you do not have visibility into them. have them check the interfaces for any errors, and also confirm speed/duplex settings. We have many customers who configure their gear with auto-neg and our CPE is hard coded. Always creates issues....food for thought
 
Thanks for your recommendations, guys!

I decided to try OSPF, and it seems to be working so far. I won't know how stable it is until early tomorrow morning when I bring in full working subnets into the mix.
 
You must log in or register to reply here.
Back
Top