Question on full redundancy

news2010a · ‎09-24-2010

Hi,

Please see sketch attached.

Imagine this is a midsize company that wants to provide full redundancy at Core, Distribution and Access Layer.

Question:
Between the two Cisco rotuers 2811A and 2811B, there is no need to have a physical connection there between the two, right? I hope not because there is no Ehternet interfaces available.

Someone suggested HSRP between the two routers... but if they are running OSPF (to handle internal routing) and BGP with service provider, I thought about setting up iBGP between the two.

Jon Marshall · ‎09-24-2010

news2010a wrote:

Hi,

Please see sketch attached.

Imagine this is a midsize company that wants to provide full redundancy at Core, Distribution and Access Layer.

Question:
Between the two Cisco rotuers 2811A and 2811B, there is no need to have a physical connection there between the two, right? I hope not because there is no Ehternet interfaces available.

Someone suggested HSRP between the two routers... but if they are running OSPF (to handle internal routing) and BGP with service provider, I thought about setting up iBGP between the two.

Marlon

First thing to say is that if this is a midsize company then why are the 3750 switches only acting as L2 switches ? Where are the vlans being routed ? For a midzised company it wouldn't make sense to route the vlans off the 2811s with router on a stick or have one flat vlan for the internal network so the setup is a but confusing from that respect.

As for the 2811 routers it really depends on what you are trying to do.

Do you want to use one ISP for all traffic and only use the other connection if the first fails or do you want to use both links. That would have a bearing on whether HSRP would be a solution for you or not.

Jon

news2010a · ‎09-24-2010

I should have said this is actually a small shop, only 50 users but we have need for high availability.

Hey, in this case, the provider is offering IP Centrex Solution. So in reality today there is only RouterA in place and all Call Functions, etc and DHCP is provided by server provider side.

Recently circuit from ISP1 failed.

So we are looking into ways to get the 2811 RouterB in place and connect that to a different provider, ISP2.

That said, let's say we want to use both ISP1 and ISP2, if possible load balance between the two. Please let me know your thoughts.

Tharak Abraham · ‎09-24-2010

Few thoughts here:-

When ISP 1 provides the DHCP for the entire infrastructure and it fails, will ISP 2 allow the scope to be reached across the internet ?

I mean will ISP 2 allow to punch a hole like that ?

The design i would add is not so pleasing if you look for redundancy...

Its better to keep the 35xx do layer 3 and intervlan routing.

This can help you in the future too since its at the distribution layer.

Since you have a smaller enterprise you may have to take some calls on improving routing efficiency

1.You can use Ospf at the core with E1 defaults downstream, not recommended but still would work

2.BGP and partial routes acceptance from both the ISP's. This is something you might wanna think ?

You would have to advertise the internal scope to both the ISP's and make sure they have it in their BGP tables.

Although, Primary and Secondary concepts of ISP's would scale well here.

Load balancing may not be possible though or maybe am wrong about that.