Why /30 for the 4 links not /28 ???

[Ibrahim Jamil]

Hi Friends pls Find attachment Why /30 for the 4 links not /28 ??? pls clear thanks

[Giuseppe Larosa]

Hello Ibrahim,

they are four point to point networks with eBGP sessions over them.

the use of four /30 is needed in order to build the eBGP sessions over each link.

If using a /28 it would not possible to have an EBGP session over each physical link.

A single IP address on each VSS would be possible and only one eBGP session would be built and you would need to use SVI interfaces (interface Vlan).

In other words with this design if one link fails the corresponding eBGP session is turned down as soon as the link is detected down.

Edit:

I am not sure you have two VSS pairs of devices, however the reasons are still the same the use of /30 anchors each eBGP session to a single physical link with no change to L2 failover over another link in the same Vlan.

Hope to help

Giuseppe

[Rolando Valenzuela]

My two cents:

It is a design alternative on how simple (or not) you want your BGP to be but please notice that you cannot define the same subnet on all four interfaces since that will create an overlap.

The only way to have them all working together will be:

• Have four different trunks (Layer 2) and the configure the IP on a SVI (Layer 3) like Giuseppe mentioned.
• Another option is combining them on a Layer 3 PortChannel

For either case you only need a total two IPs, therefore one /30 should work.

There are some benefits of using four /30:

• P2P
• Everything is directly connected
• Better handling of link failure
• No STP
• No load balacing (which during a link failure may force you to account for hop count and delays)
• No single point of failure by having just one BGP session

Regards.

Rolando A. Valenzuela.

[Ibrahim Jamil]

Hi Giuseppe , Hi Rolando

What if we have OSPF ,is it recomended to have /30 or its okay with /30

Any Design issue with regard to routing and fast convergence

thanks

[Giuseppe Larosa]

Hello Ibrahim,

also if using OSPF the use of point to point routed links provides the same benefits as noted also by Rolando: there is no STP involved. If one link fails it is managed at OSI layer3 by OSPF.

For fast convergence you can use BFD that is well suited for point to point links.

Edit:

if using point to point links you can save on DR/BDR election by using

ip ospf network point-to-point

on each link. This skips all the DR/BDR election that would occur on a broadcast network.

This is specific of OSPF.

Hope to help

Giuseppe

[Ibrahim Jamil]

Okay freinds

thanks