Running routing protocol over p2p vs. VLAN interface

Unanswered Question
Jan 28th, 2008

This is more a design question. We're thinking about changing our campus core from layer 2 to layer 3. What are the advantages to running a routing protocol over a point to point link instead of a VLAN interface? I've read that point-to-point links provide the fastest convergence for routing protocols.

My understanding to this point is when p2p links drop all routes and neighbors associated with that link are dropped. With a VLAN interface the underlying protocol has no way of signaling to the routing protocol that interface state changed. With this design OSPF timers are responsible for killing OSPF session. During this period traffic is black holed.

Is this right? I guess I'm really looking for validation...

Thanks,

Ryan

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (2 ratings)
Loading.
Edison Ortiz Mon, 01/28/2008 - 09:09

There is some validity into that. You may also add that you eliminate Spanning-Tree on the switchport when changing from L2 to L3 links.

__

Edison.

dnewell24 Mon, 01/28/2008 - 09:25

ediortiz currently our L2 core is a loop-free topology. Other than the overhead STP is really not an issue. When you say some validity I assume not everything I say is right. Can you please shed some light?

Thanks,

Ryan

Edison Ortiz Mon, 01/28/2008 - 10:17

Even on a loop-free topology, the switchport does not go into forwarding mode right away when acting as Layer2. There is some delay, do you want to test?

Type show spanning-tree interface x/x after issuing 'no shut' on a port.

I'll have to test on your proposal before giving it a full validity :) It has some merit , though.

__

Edison.

Jon Marshall Mon, 01/28/2008 - 10:21

Hi Ryan

With a P2P routed link the routing protocol still has to miss a number of keepalives before it realises it's neighbor has gone down. Not sure what you mean by underlying protocol as i assume we are talking ethernet in both cases ? as otherwise it would not be a straight comparison.

I believe, (Edison please feel free to correct), that where you get a benefit is if the link comes back up. With L2 STP must do it's convergence before the routing protocol can form a neighborship, whereas with L3 you do not have to wait on STP.

Jon

Edison Ortiz Mon, 01/28/2008 - 10:30

> With L2 STP must do it's convergence before the routing protocol can form a neighborship,

> whereas with L3 you do not have to wait on STP.

Yep

Edison Ortiz Mon, 01/28/2008 - 10:32

> With a P2P routed link the routing protocol still has to miss a number of keepalives before it realises

> it's neighbor has gone down.

He does bring up a very valid point though. The interface state change should trigger the routing protocol faster on the L3 port, but that's something that need to be tested. If Ryan has some spare gear, he can do us the favor :)

dnewell24 Mon, 01/28/2008 - 11:50

I was looking to you smart guys for the answer!! If only I had test gear...

High Availability Campus Recovery Analysis

http://www.cisco.com/en/US/solutions/ns340/ns414/ns742/ns656/net_design_guidance0900aecd801a89fc.pdf

Pulled from page 9 -

In a Layer 3 core design, convergence times for traffic flowing from any distribution switch to any other

distribution switch are primarily dependent on the detection of link loss on the distribution switches. On

GigE and 10GigE fiber, link loss detection is normally accomplished using the Remote Fault detection

mechanism implemented as a part of the 802.3z and 802.3ae link negotiation protocols.

Note Please see IEEE standards 802.3ae & 802.3z for details on the remote fault operation for 10GigE and

GigE respectively

Once the distribution switch detects link loss, it processes a link down event that triggers the following

three-step process:

1. Removal of the entries in the routing table associated with the failed link

2. Update of the software Cisco Express Forwarding (CEF) table to reflect the loss of the next hop

adjacencies for those routes affected.

3. Update of the hardware tables to reflect the change in the valid next hop adjacencies contained in

the software table.

My take -

Assuming equal cost redundant links exist to the core p2p links provide faster convergence over peering across VLAN interfaces. When the p2p link fails all associated routes are removed leaving only the other redundant path. The routing peer hasn't timed out yet but the routes have been removed from the table because the link failed.

Make senses to me...

Jon Marshall Mon, 01/28/2008 - 11:54

"I was looking to you smart guys for the answer!! "

Edison - he has to be referring to you because i'm not that smart :)

Aside from that. You didn't mention you had redundant equal cost paths in your scenario. I have tested this before ie. redundant links with STP, RSTP and EIGRP. EIGRP won every time because as you say you still have an immediate path with the L3 routed links, at most you would lose one packet but i very rarely did in my tests.

Jon

Edison Ortiz Tue, 01/29/2008 - 18:33

> Edison - he has to be referring to you because i'm not that smart :)

Right ............. :|

Actions

This Discussion