We're now going to look at the next
example. We're scaling up the previous
one where the network is connecting to
the upstream provider and several local
peers through an internet exchange point.
This again is a very common situation in
many regions of the internet--you connect
to your transit provider to see the
whole internet and you connect to
the local internet exchange point
so that local traffic stays local
for the same reasons as earlier.
If we look at the diagram we
now see a s 100 rata a is connecting to
an Ethernet switch at the internet
exchange point and the several other
networks present there Roger C is
connecting to s 130 the upstream
provider so how do we configure all of
this we're going to announce the slash
19-acre gate to every neighboring a s
that's before we will accept the default
route only from the upstream provider we
don't need the foo BGP table and we
accept all rights which are originated
by the exchange point peers if we look
at the router a configuration we can
scale the example we saw earlier we can
set up a peer group we call it I X PPS
and in that we put in our ID bond prefix
list that lets our address block out we
strip out private e asses we could send
the BGP community attribute if we
require that and we can also set up a
road map to set local preference for the
prefixes that we hear from the exchange
point peers this is actually a very
common type of peer group used for
operators peering at an internet
exchange point and then we apply this
peer group to the exchange point
neighbors I've got some examples here on
the slide each neighbor has the peer
group applied to it we've got an inbound
filter because we always filter
bgp sessions in and out as we have
learned before and the prefix list is
set up for each peer the right map set
local pref basically everything that we
hear from the exchange point peer we set
local preference high with made at 150
in this example I realize right now we
don't need this because we're only here
the default right from the upstream
provider and the prefixes we hear from
exchange point peers are more specific
and therefore better paths however in
future we may get a second upstream
provider we may get prefixes learned
from the two up streams and we may be a
prior preferences there so we're
starting to introduce the concept of
applying local policy to prefixes
depending on where the heard from so
I've set local preference 150 for
exchange point peers and that means
there will be higher priority than say
prefixes we learn from any other source
note that router a does not generate to
aggregate for a s 100 if router a
becomes disconnected from the backbone
then the aggregate would simply
disappear from the exchange part and BGP
failover would work as expected if
router a did generate the aggregate and
then if router a was disconnected from
the backbone it would carry on
announcing the aggregate to the exchange
point what would happen is outbound
traffic will go to the upstream provider
it would find its way to one of the
peers as expected but then the peer
would see the best path back to you via
your router a which is announcing the
aggregator at the exchange point and
because it has nowhere for the traffic
to go it would simply be dropped in the
null interface or discard interface on
the router so you'd end up cutting
yourself off from the local peers so the
best policy and the best practice if the
router is remote from your network
do not generate the aggregate on that
router secondly notice the inbound right
map which sets the local preference
higher than the default and it's a
visual reminder that BGP best path the
local traffic will be across the
exchange point and as I mentioned
earlier allows the future case for
operator may take bgp rights from their
upstream providers if we look around to
see configuration this is the one
connecting to the upstream we simply
allow the default in and send an
aggregate out as we did before and just
to conclude notice the router a
configuration the prefix list is higher
maintenance but it's safer if a peer is
it going to introduce a new prefix they
will let all members of exchange point
know that they have a new address block
they want to introduce and so everybody
can update their filters and we don't
generate the aggregate for s 100 on
router a either the exchange point
traffic goes to and from the local ixb
everything else goes to the upstream
notes for the I XP peers that the
peering ISPs the exchange point will
only exchange prefixes they originate
and sometimes they exchange prefixes
from neighboring guesses as well usually
their customers so as before be aware
that the exchange by border router
should carry only the prefixes you want
the exchange point peers to receive and
the destinations you want them to be
able to reach if you got the foo BGP
table there or if you have a default
route there your peers could potentially
transit your backbone to the rest of the
internet also if the exchange point
routers that the i-x are distant from
your backbone it's really important not
to originate your address block
at the exchange point router.
© Produced by Philip Smith and the Network Startup Resource Center, through the University of Oregon.
Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)
This is a human-readable summary of (and not a substitute for) the license. Disclaimer. You are free to: Share — copy and redistribute the material in any medium or format Adapt — remix, transform, and build upon the material The licensor cannot revoke these freedoms as long as you follow the license terms. Under the following terms: Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use. NonCommercial — You may not use the material for commercial purposes. No additional restrictions — You may not apply legal terms or technological measures that legally restrict others from doing anything the license permits.