I'm trying to provide redundant default gateways for 2950 L2 switches that daisy-chained between L3 3550 switches as shown in the attached drawing. All connections are fiber, and there's no opportunity to home run all the L2 switches to the L3 gateways with new fiber connections. I'd like to provide fail-over if the fiber link between Switch A and Switch B is cut (this would also cut the gig link between Node 1 and 2). So lets say that both Switch A and Switch B are using Node 1 as a default gateway, the cable gets cut, HSRP determines that Node 2 is the new active gateway, but now Switch A can't reach it because it's on the wrong side of the cut. I thought about using GLBP to load balance between the two nodes, but if the cable gets cut would half the packets be lost? Is it possible to do this with spanning tree? Other ideas?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations gkittelson on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
HSRP Router Redundancy
- Thread starter rbautch
- Start date
![[americanflag]](/data/assets/smilies/americanflag.gif "[americanflag] [americanflag]")
Go Army!Actually if you get a cut between those switches hsrp will think both sides are active because there is no path for hsrp to talk between those 3550's so both sides will go active. I think it would still work . If those 2 2950's are close together then just run another cable and make that link a ehterchannel , it is highly unlikely you are going to ever lose 2 connections .
- Thread starter
- #4
Thanks for the responses. Actually, there's about 16 switches between nodes, spaced about a mile apart. I only showed two of them in the diagram for simplicity. The switches are connected with a fiber optic cable buried in the ground, and I want to guard against a full cable cut, which would sever the link between the switches and the gig link that directly connects the nodes.
- Thread starter
- #5
I'm going to test it using spanning tree. I'm not sure it will work because the interface on the node switch that serves as the gateway will be different depending on which leg of the ring traffic is approaching from. Instead of assigning the gateway IP to the interface on the node switch, perhaps switched virtual interfaces would solve this?
- Status