02-20-2010 01:18 AM - edited 03-06-2019 09:48 AM
I'm looking for best practice, that would optimize the network design.
currently, all Distrubution switches are connected to Core switches (dual core design) by trunks.
We are going to more these trunks to routed ports. There are two ways of doing it.
1-use /30 subnets to create interfaces on both Distrubution and core switch. this will involve the core switch in routing decisions.
2-use single subnet to connect various Distrubution switches to each other (for redudency two subnet can be used). this will make the core switch act as a switch only and will not involve in routing decision.
I was wondering which practice would be best.
Solved! Go to Solution.
02-20-2010 07:53 AM
CSCO11584685 wrote:
just last question.
How do they use /31. Since it will give only two IP addresses. one is network, second is broadcast. Cisco router/switch will warn that /31 should be used only on point to point links.
I would like to use /31 since it will have lesser number of ips wasted on broadcast or network. but again, how does it work.
Cisco supports /31 on point to point links and they simply treat both available addresses as host addresses -
http://www.cisco.com/en/US/docs/ios/12_2t/12_2t2/feature/guide/ft31addr.html
If you are using private addressing internally i could never reallly see the point myself as you can always allocate some more private addressing. It's more useful for conserving public addressing.
Jon
02-20-2010 02:28 AM
CSCO11584685 wrote:
I'm looking for best practice, that would optimize the network design.
currently, all Distrubution switches are connected to Core switches (dual core design) by trunks.
We are going to more these trunks to routed ports. There are two ways of doing it.
1-use /30 subnets to create interfaces on both Distrubution and core switch. this will involve the core switch in routing decisions.
2-use single subnet to connect various Distrubution switches to each other (for redudency two subnet can be used). this will make the core switch act as a switch only and will not involve in routing decision.
I was wondering which practice would be best.
Best pratice depends on what else you have going on in your network. Cisco used to recommend L2 in the core because it was a lot faster than L3 but nowadays with L3 switching there is little difference and Cisco recommendation is to use L3 in the core.
Personally i would go with L3 as long as it fits your setup ie. do you need to span vlans across the core between distribution switches because if you do then you cannot use routed links from the distro to core. But if you can contain vlans within a pair of distro switches assuming you have pairs of distro switches interconnected then L3 is the way to go. It stops STP going from the distro switches to the core switches and therefore isolates broadcast domains to the distro and access-layer switches.
So option 1 would be the one i would choose but as i say you need to be sure that you do not need any L2 in your core before you do this.
Jon
02-20-2010 02:41 AM
I think im not very clear. or the scenario is not clear. let me try to draw it
Dist Switch1 ----> Core <----- Dist switch 2.
The link will be L3. it is L2 right now. we are moving it to L3, cause we want to stop all the broadcast, etc. since the vlans are not interconnected through the dist switches.
My question is. should i do like:
Dist switch 1 (10.0.0.1/30) ------> (10.0.0.2/30) Core (10.0.0.5/30) <------ (10.0.0.6/30) Dist switch 2
OR
Dist switch 1 (10.0.0.1/24) ------> (no ip, switchport) Core (no ip, switchport) <------ (10.0.0.2/24) Dist switch 2
As i said, im looking for the recommended practice. I just read somewhere (http://www.techexams.net/forums/ccnp/49215-distribution-core-switch-connections.html) about /31 addressed
CORE(config-if)#ip add 192.168.2.1 255.255.255.254
% Warning: use /31 mask on non point-to-point interface cautiously
*interface GigabitEthernet0/11
no switchport
ip address 192.168.2.1 255.255.255.254*
02-20-2010 03:11 AM
CSCO11584685 wrote:
I think im not very clear. or the scenario is not clear. let me try to draw it
Dist Switch1 ----> Core <----- Dist switch 2.
The link will be L3. it is L2 right now. we are moving it to L3, cause we want to stop all the broadcast, etc. since the vlans are not interconnected through the dist switches.
My question is. should i do like:
Dist switch 1 (10.0.0.1/30) ------> (10.0.0.2/30) Core (10.0.0.5/30) <------ (10.0.0.6/30) Dist switch 2
OR
Dist switch 1 (10.0.0.1/24) ------> (no ip, switchport) Core (no ip, switchport) <------ (10.0.0.2/24) Dist switch 2
As i said, im looking for the recommended practice. I just read somewhere (http://www.techexams.net/forums/ccnp/49215-distribution-core-switch-connections.html) about /31 addressed
Ah okay, i understand.
I would still go with /30 routed P2P links because in the 2nd example you are still actually extending L2 through the core. I can't see the benefit you gain by doing this either.
In addtion to which with option 2 if dist switch 2 fails then dist switch 1 is not immediately aware of it because it's connection to the core is still up so you are reliant on routing protocol timers. The core switch is aware of the dist2 falure but it is not involved in the routing neigborships so there is nothing it can do.
With option 1 this is not the same. If dist2 fails the core switch is immediately aware of it because it has a P2P link and is a L3 peer so it can immediately notify dist1 by withdrawing routes.
Jon
02-20-2010 04:12 AM
just last question.
How do they use /31. Since it will give only two IP addresses. one is network, second is broadcast. Cisco router/switch will warn that /31 should be used only on point to point links.
I would like to use /31 since it will have lesser number of ips wasted on broadcast or network. but again, how does it work.
02-20-2010 07:53 AM
CSCO11584685 wrote:
just last question.
How do they use /31. Since it will give only two IP addresses. one is network, second is broadcast. Cisco router/switch will warn that /31 should be used only on point to point links.
I would like to use /31 since it will have lesser number of ips wasted on broadcast or network. but again, how does it work.
Cisco supports /31 on point to point links and they simply treat both available addresses as host addresses -
http://www.cisco.com/en/US/docs/ios/12_2t/12_2t2/feature/guide/ft31addr.html
If you are using private addressing internally i could never reallly see the point myself as you can always allocate some more private addressing. It's more useful for conserving public addressing.
Jon
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide