Quanta Cloud Technology QuantaMesh QNOS5 Configuration Manual Download Page 199

Page: 199 / 256

199

7.5.2.

BGP Behavior

To begin with, BGP systems form a TCP/IP connection between one another to exchange NLRIs. First, they
exchange messages to open and confirm the connection parameters. The initial data flow is the entire BGP
routing table. Incremental updates are sent as the routing tables change. BGP does not require periodic
refresh of the entire BGP routing table because it relies on the reliable transport provided by TCP. Therefore,
a BGP speaker must retain the current version of the entire BGP routing tables of all of its peers for the
duration of the connection. Keepalive messages are sent periodically to ensure that connection is active.
Notification messages are sent in response to errors or special conditions. If a connection encounters an
error condition, a notification message is sent and the connection is closed.

Routes are advertised between a pair of BGP speakers in UPDATE messages, where the network destinations
are the systems whose IP addresses are reported in the NLRI field, and the AS path for those destinations is
part of the information reported in the path attributes fields of the same UPDATE message, along with
various other BGP attributes. Routes are stored in local Routing Information Bases (RIBs). Logically, all routes
learned from a particular BGP peer are kept in a local Adj-RIB-In, and all routes learned from all BGP peers
are held in a Loc-RIB, which serves as the central database for BGP to determine the

best

path to a

particular network destination. Additionally, local policy configuration may filter or modify the BGP
attributes of NLRIs that are received from BGP peers.

Once BGP has chosen the

best

path to a network destination based on the BGP attributes given in an NLRI

(also known as the

decision process

), it must determine if there is connectivity to the destination defined by

the BGP

nexthop

attribute from the

best

NLRI. Here, BGP performs

nexthop resolution

by referencing the

local router's forwarding table, which is populated with routes installed by IGP protocols. If connectivity to
the BGP nexthop is found (i.e. resolved), then the corresponding BGP route can be installed to the local
router's forwarding table, using the

real

nexthop information from the IGP route that was used to resolve

the BGP nexthop.

Finally, BGP routes that have been installed in the local router's forwarding table are eligible to be advertised
to connected BGP peers. BGP advertises these routes to each connected peer, typically resetting the BGP
nexthop attribute to be the local IP address for the BGP peer connection. Additionally, local policy
configuration may filter or modify the NLRIs that are advertised to these BGP peers.

For a more detailed and comprehensive description of BGP protocol behavior, refer to the BGP-4 Protocol
Specification (RFC1771/draft-ietf-idr-bgp4-26).

7.5.2.1.

BGP Route Selection

QNOS BGP uses the following route selection rules:

Prefer the route with the higher local preference

Prefer a locally-originated route over a non-locally originated route

Prefer the route with the shorter AS Path

Prefer the route with the lower ORIGIN. IGP is better than EGP is better than INCOMPLETE.