Shulou(Shulou.com)06/01 Report--

The reasons why fabric in large data centers tend to use BGP as an internal routing protocol:

1.BGP has simpler state machines and databases than IGP.

2.BGP exchanges routes, direct commands, and does not need to run algorithms (such as SPF) to calculate the path

3.BGP supports hop-by-hop traffic engineering

4. In terms of route propagation, the BGP range is smaller (peer propagation, OSPF requires link-state flooding) and is therefore more stable.

Choice of IBGP and EBGP:

1.EBGP sessions are generally established on the link of the BGP peer, and IBGP is generally established on the loopback interface of the BGP peer

2.EBGP uses AS-path to prevent loops. IBGP uses the split horizon principle within the same AS: routes received from EBGP peers are sent to other EBGP and IGBP peers by default, and routes received from IBGP peers are advertised to EBGP peers by default but not to other IBGP peers. Due to the split horizon principle, IBGP generally uses reflectors (route reflector) or confederation, and generally uses RR.

3. Therefore, EBGP is generally used, which is simple and straightforward.

When EBGP applies data center fabric:

1. Prevent loops through the AS-path attribute

two。 Next-hop-self is set by default for advertised routes

3. Automatically readvertise routes learned from other EBGP sessions

Of course, IBGP can also be used for data center fabric, which generally mimics EBGP behavior:

1. Set up RR and use originator ID (originator-ID) and cluster list (cluster-list) to prevent loops

two。 To readvertise a route is to set the next-hop-self.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.