At 6:32 PM -0400 5/10/02, dre wrote:
>""Chris Headings""  wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
>>  We are an ISP in So. Cal.  We are gearing up to open other offices in
>>  bewteen Arizona and Ca...
>>
>>  We are trying to decide what would be the best way of intergrating our
ISP
>>  network...like...
>>
>>  Should we just continue to use our one ASN and have all traffic come back
>to
>>  the Global NOC, or should get new ASN's for each location???
>
>Philip Smith's Multihoming NANOG presentation covers disconnected backbones
>and ASN usage.

There's also the "IOS Essentials Every ISP Should Know," the URL for 
which I don't have handy but is at Cisco, as well as the ISP Workshop 
series.

>
>>  Should we be using IS-IS, IBGP/EBGP, OSPF as the protocol to link all
>remote
>>  locations???
>
>Yes.  If you can afford it (and I can't see why you wouldn't be able to),
>you should
>have all routers take part in IBGP.

Just to clarify something I'm sure dre well knows, having all routers 
participate in iBGP doesn't mean they have to be in a flat topology. 
The rough rule of thumb has long been you don't want more than 20-30 
BGP sessions of any type on one physical router, which is the 
motivation for iBGP scalability techniques such as confederations and 
route reflectors.  Of course, this is a worst-case number, tending to 
assume full routes and a lot of churn.  You could probably get away 
with a good deal more if you're just having eBGP connections to 
customers advertising their own routes and accepting default.

Some of the scalability issues are in the IETF BMWG drafts 
(http://www.ietf.org/html.charters/bmwg-charter.html).  We've done 
eBGP first, but the terminology draft is relevant to both eBGP and 
iBGP, and the appendix to the eBGP methodology draft gives some 
typical sizing.

>Doesn't matter about the IGP, some
>people
>prefer IS-IS, and others prefer OSPF.  GlobalCrossing and AboveNet (and
>probably
>other ISP's) had to move to IS-IS eventually, and migration is terrible.  If
>you have
>the knowledge or time to invest in IS-IS, I would suggest looking into that
>as an option
>before your network grows too large.  Fortunately, there are a lot of really
>good
>resources out there today (for both IS-IS and OSPF, actually ;> ).
>
>>  Obviously cost in an issue, but throwing cost out the door, what is the
>>  ideal way of linking ALL offices, using a good level of redundancy and
>great
>>  preformance...
>
>Metro Ethernet and wavelength services make purchasing distance bandwidth a
>lot
>easier.  Co-location has also come a long way, making transit and peering
>very easy
>and cost effective.  You are no longer being forced to connect into some
>strange,
>unknown remote POP via costly (and difficult to provision) SONET circuits
>and router
>interfaces.  Look into your Exchange Points and Metro Providers (CLEC's for
>fiber and
>optical wavelengths) possibilities.

Very valid points.  There's a lot of discussion of exchange points at 
www.ripe.net.  Exchange points operated by many organizations are 
more common in Europe than in the US, and indeed there's now a 
European Exchange Operators Forum (or something along those lines -- 
you can find it at RIPE).

When doing these things, also think about local loop diversity.
-- 
"What Problem are you trying to solve?"
***send Cisco questions to the list, so all can benefit -- not 
directly to me***
********************************************************************************
Howard C. Berkowitz      [EMAIL PROTECTED]
Chief Technology Officer, GettLab/Gett Communications http://www.gettlabs.com
Technical Director, CertificationZone.com http://www.certificationzone.com
"retired" Certified Cisco Systems Instructor (CID) #93005




Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=43916&t=43836
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]

Reply via email to