Jump to content
Hamer Fan Club Message Center
  • 0

Any Route/Switch Wizards? Help Needed.


Lockbody

Question

Posted

I'm more of a voice guy, but I've got a network to reconfigure, and I'm not sure I'm going about the best way. The HFC is a treasure trove of knowledge, so I'm going to throw this out there and see if anybody can help.

I've got a customer with two MetroE connections to the same host, but only one ME at the remotes, all over the same vlan provided from Telco. In the past it was setup:

Host:

7200 router

ME1-172.16.0.3/25 - Dot1q vlan 278 int f1/0

ME2-172.16.0.129/25 - Dot1q vlan 278 int f2/0

(these come in from two differnet COs in case one fails. Both are always active)

Remote ex:

3750 switch

int vlan 278

172.16.0.9

172.16.0.135 secondary

vlan is applied to one int on sw

BGP configured on host and remotes controlled traffic with peer and peerback statements to allow remotes to reach host in case one CO fails, and to allow simple load balancing.

All was fine and dandy until the 7200 router at the host was replaced with a 4503 switch.

assumptions:

There are two MetroE circuits termininating into the host 4503 both must be up, with some sort of load balancing.

All traffic must enter the cloud tagged vlan 278

Routed/tagged interfaces cannot be created on the 4503, only switchports

bgp is not needed

My first though was to have a secondary ip on on the int vlan 278 like all of the remotes, but bgp doesn't know how to deal with two active paths with two different ips and will almost immediately start de-neighboring the the remotes until one interface is shutdown.

I'm thinking of getting rid of BGP altogether along with the secondary interfaces and moving to a simpler routing protocol like EIGRP. So my basic config would be:

Host:

int vlan 278

172.16.0.3/24

int g1/0/1 access vlan 278

int g1/0/2 access vlan 278

eigrp 1

network 172.16.0.0

network 192.168.1.0 (local)

remote

int vlan 278

172.16.0.9

int g1/0/1 access vlan 278

eigrp 1

network 172.16.0.0

network 192.168.2.0

So, I guess I'm wondering if this is the best way to do it. Will the traffic still go over both int at the host to the remotes with some sort of load balancing between the two circuits?

Or

Should I leave the secondary IPs configured at the remotes and use something like EIGRP variance command to load balance?

Any help would be appreciated. Thanks.

3 answers to this question

Recommended Posts

Posted

I'm not familiar with the 4503 line, since it's not something we have in my environment (6500's, 7200, 2800, 3800 Routers & Nexus mostly). Been a long week and I'm a little mushy, but I think I'm missing something in your scenario. You're saying that your 7200 router's been replaced with a 4503 switch. And that you can't do routed/dot1q tagging on the 4503 interfaces. But I also think I'm reading in your assumptions that you're terminating your 2 metro-E circuits on the 4503.

So where are you doing your routing? :lol:

Depending on your access to the 4503 can you do a show version as well as a show module? Curious to see what IOS/feature set you have as well as what type of Supervisor engine and line cards.

Posted

So where are you doing your routing? :lol:

The magic happens in the cloud, and that's part of my problem.

Turns out that the problem is due to the particular service this customer has. Most of my dealings with MetroE in the past let me do whatever and hand whatever into the cloud, do my own end-to-end QOS, etc...

Not this offering. It comes down to that one switch will not be able to handle both ME circuits. My boss and I sat down and tried every thing we could think of and nothing we did would allow both of those circuits to route traffic (L1 and L2 was up, but no L3, pings etc...). Shut one down and ping city.

I think the solution will be to add another switch like a 3750 to terminate the second circuit and HSRP the two switches.

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...