Seems rather presumptuous of Cisco to speak for every ISP in the world....
In order to limit the number of routes being advertised on the internet, I
believe it was considered "best current practice" to limit prefix length to
/19 or shorter. ( can't find the RFC at the moment, but I recall it being
referenced several times in various threads on the NANOG list. )
Obviously, with well over 100K routes in "the internet routing table" there
are a great number of longer prefixes being advertised, no doubt in great
part because of the number of companies that are "connected to multiple
ISP's so they can load balance across the internet"
Prefix advertising my be influenced by peering arrangements, downstream and
upstream agreements, and customer requirements. Generally, once holes are
punched through CIDR blocks, what can anyone do?
When someone makes a statement like you attribute to Cisco, one must always
follow up with specifics to determine what is really meant. Not all routes
seen in one provider's network routing tables are necessarily present in the
tables of another provider.
Chuck
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]] On Behalf Of
Murphy, Brennan
Sent: Monday, May 07, 2001 4:01 PM
To: '[EMAIL PROTECTED]'; Charlie Winckless; Murphy, Brennan
Cc: [EMAIL PROTECTED]
Subject: RE: real world BGP question
Cisco told me today that a /24 drawn from Class C space
has a better chance of being propogated throughout the Internet
than a /24 taken from Class B space. Anyone disagree with that?
Can anyone recommend a good source of info on this. Ive checked
Halabi.
I came across a good reference during my quest www.traceroute.org
Unfortunately, it doesnt offer plain answers to my questions.
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]
Sent: Friday, April 27, 2001 1:58 PM
To: Charlie Winckless; 'Murphy, Brennan'
Cc: [EMAIL PROTECTED]
Subject: Re: real world BGP question
Currently on a US basis a /24 would generaly work. Internationaly (Europe)
most providers would filter out anywhing longer then /20.
----- Original Message -----
From: "Charlie Winckless"
To: "'Murphy, Brennan'"
Cc:
Sent: Friday, April 27, 2001 1:18 PM
Subject: RE: real world BGP question
> I used to work for VERIO. At that time they would not
> router smaller than /19 on their backbone.
>
> This may have changed.
>
> > -----Original Message-----
> > From: Murphy, Brennan [mailto:[EMAIL PROTECTED]]
> > Sent: Friday, April 27, 2001 10:46 AM
> > To: 'Michelle T'; '[EMAIL PROTECTED]'
> > Subject: RE: real world BGP question
> >
> >
> > I guess that is my real question: what is the longest prefix that
> > is exchanged among/between major carriers.
> >
> > The real world example here is what if you had 4 server farms
> > answering
> > to one DNS name: ftp.foo.com You have Round Robin DNS running
> > round trip times to match a user with their nearest server farm....
> > so it sends back the closest/fastest IP. The question is, how
> > big do those
> > subnets for the server farms have to be in order to be maximally
> > advertised throughout the internet?
> >
> > So, I've seen two answers in this thread /20-21 or /24. I wonder
> > where I could find the real answer? Maybe Halabi has a link in the
> > back of his book to an organization that maintains info such as
> > this.
> >
> > Any more input is greatly appreciated. Thanks to all who have
> > responded.
> > I figured this question was a relavant BGP question relating
> > our studies
> > to an actual scenario.
> >
> >
> >
> > -----Original Message-----
> > From: Michelle T [mailto:[EMAIL PROTECTED]]
> > Sent: Friday, April 27, 2001 12:06 PM
> > To: Murphy, Brennan; [EMAIL PROTECTED]
> > Subject: RE: real world BGP question
> >
> >
> > /24 is the longest prefix you will see accepted by nearly any
> > carrier out
> > there. Many will only accept /20 or /21. All perform
> > aggregation to some
> > degree, though exception routing is allowed to send the /24's
> > (/23, /22,
> > etc) out to the ISP peers when the customer is multi-homed two diverse
> > carriers.
> >
> > I can tell you that I work for a Tier 1 ISP and we accept
> > longer prefixes
> > for many customers who are multi-homed just to us. They use
> > the various
> > subnets as a simple method of controlling inbound traffic
> > distribution, to
> > enact policy, etc...
> >
> > Many times we see multi-homed (dual-ISP) customers advertise
> > an aggregate
> > /16 or longer and also advertise /24's for the same reaason (policy,
> > distribution, etc).
> >
> > Michelle Truman
> >
> >
> > -----Original Message-----
> > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of
> > Murphy, Brennan
> > Sent: Friday, April 27, 2001 10:28 AM
> > To: '[EMAIL PROTECTED]'
> > Subject: real world BGP question
> >
> >
> > What is the smallest subnet that major carriers will exchange with one
> > another? /24..../26.../27?? I know that the real issue is
> > the size of
> > the route table.
> >
> > I'm just wondering about the reallity of scenarios that
> > Habali describes
> > where an institution advertises an aggregate with specific subnets.
> >
> > I know that when you're multi-homed to a carrier, that carrier will
> > sometimes
> > take your /26 and /27 nets to help route inbound traffic but
> > that carrier
> > will not advertise those nets to its neighbors.....at least
> > thats what I've
> > heard.
> >
> > Anyone have any real world experience with this? Or is there a URL
> > I could read up on?
> >
> > Thanks,
> >
> > BM
> > **Please read:http://www.groupstudy.com/list/posting.html
> > **Please read:http://www.groupstudy.com/list/posting.html
> **Please read:http://www.groupstudy.com/list/posting.html
**Please read:http://www.groupstudy.com/list/posting.html
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=3506&t=3506
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
