November 1996 - Test - lists.nanog.org

New public exchanges
by bmanning＠ISI.EDU 15 Nov '96

15 Nov '96

Some recent updates on http://info.ra.net/div7/ra/naps.html BNIX - revised Belgian exchange MAE-Dallas - another Texas, USA exchage OIX - Oregon USA exchange has some nice pages up ISPA - A front for South African exchange work KIX - A Korean Exchange -- --bill

1 0

Re: Access Lists
by Paul Ferguson 14 Nov '96

14 Nov '96

At 12:40 PM 11/14/96 -0400, Frank_Troy_at_NSD1%SED01(a)va-fch31.csc.com wrote: > > How safe is it to filter on AS numbers without using access lists? > Define: safe. What are you trying to accomplish? - paul

1 0

Re: high packet loss within Sprint infrastructure?
by Paul J. Zawada 14 Nov '96

14 Nov '96

I guess you haven't read this list very long... The NSPs have no packet loss and the problem is obviously that the end sites aren't willing to pay for large enough pipes. --zawada [being facetiously sarcastic] At 02:44 PM 11/13/96 -0800, you wrote: >I hope this is an appropriate question for this list... > >We have been experiencing severe packet loss, apparently within the Sprint >infrastructure, for several weeks. Recently we've seen as high as 20% >loss. Our users are screaming. They can't keep sessions open to resource >providers, etc., to whom they are paying large annual fees. Not at all >good. > >We have been receiving conflicting stories fron Sprint about where the >problem(s) is(are), who is responsible, what is being done (if anything), >etc. > >The loss appears to us to be worst at MAE-WEST but it seems to be happening >more broadly as well. > >Can anyone else out there help shed light on this? Is anyone else aware of >similar behavior? > >Thanks, > David Wasley > University of California > > > >

1 0

Access Lists
by Frank_Troy_at_NSD1%SED01＠va-fch31.csc.com 14 Nov '96

14 Nov '96

How safe is it to filter on AS numbers without using access lists?

1 0

Re: high packet loss within Sprint infrastructure?
by Stephen Meuse 13 Nov '96

13 Nov '96

On Nov 13, 2:44pm, David L. Wasley wrote: > Subject: high packet loss within Sprint infrastructure? > I hope this is an appropriate question for this list... > > We have been experiencing severe packet loss, apparently within the Sprint > infrastructure, for several weeks. Recently we've seen as high as 20% > loss. Our users are screaming. They can't keep sessions open to resource > providers, etc., to whom they are paying large annual fees. Not at all > good. I've heard recently that sprint will be adding more capacity to their Mae-West connection, but at this point it is just a rumor.... > > We have been receiving conflicting stories fron Sprint about where the > problem(s) is(are), who is responsible, what is being done (if anything), > etc. > > The loss appears to us to be worst at MAE-WEST but it seems to be happening > more broadly as well. > > Can anyone else out there help shed light on this? Is anyone else aware of > similar behavior? > > Thanks, > David Wasley > University of California > > >-- End of excerpt from David L. Wasley -- ------------------------------------------------------------------------------- Stephen Meuse ___ ___ _ __ ____ __ __ BBN Planet Corporation / _ )/ _ )/ |/ / / __ \/ /___ _____ ___ / /_ Network Analyst / _ / _ / / / /_/ / / __ `/ __ \/ _ \/ __/ 150 Cambridge Park Drive /____/____/_/|_/ / ____/ / /_/ / / / / __/ /_ Cambridge, MA. 02140 /_/ /_/\__,_/_/ /_/\___/\__/ ------------------------------------------------------------------------------- Who is this "cron" person? And why is he mailing me?!! -------------------------------------------------------------------------------

2 1

Re: Why doesn't BGP... -Reply
by Vadim Antonov 13 Nov '96

13 Nov '96

Neal Castagnoli <Neal_Castagnoli(a)novell.com> wrote: >Routing does not use an exponential NP-Complete algorithm (like the >travelling salesman). The travelling salesman problem tries to solve the >cheapest way in which a salesman can visit every one of a set of cities. >In fact, routing can be done in order (n) with a bounded metric and >bounded distance, since it really is only trying to find the cheapest way >to get to a single destination. The routing mentioned was optimal routing of virtual circuits. It thought of as NP-complete. There's quite a lot of science (commodity flows, and such) essentially investigating efficient ways to find quasi-optimal solutions for the problem. >What I don't know, is why is it that SS7, the telephone routing protocol, >can do some of the things that are required, like load sharing across >unequal paths, for example. Does anyone have any insight into this? They simply precompute routing for different scenarios and then just load it. Precomputation can take days at pretty big machines. SS7 is no magic -- just a protocol. --vadim

1 0

high packet loss within Sprint infrastructure?
by David L. Wasley 13 Nov '96

13 Nov '96

I hope this is an appropriate question for this list... We have been experiencing severe packet loss, apparently within the Sprint infrastructure, for several weeks. Recently we've seen as high as 20% loss. Our users are screaming. They can't keep sessions open to resource providers, etc., to whom they are paying large annual fees. Not at all good. We have been receiving conflicting stories fron Sprint about where the problem(s) is(are), who is responsible, what is being done (if anything), etc. The loss appears to us to be worst at MAE-WEST but it seems to be happening more broadly as well. Can anyone else out there help shed light on this? Is anyone else aware of similar behavior? Thanks, David Wasley University of California

2 1

"Naive" (not so stupid) Newbie bashfully seeks help
by Erik Britt-Webb 13 Nov '96

13 Nov '96

Ok, I wont' approach the mailbomb subject. I've been on the list for a week, but haven't seen anything applying to my ignorance gap--which is considerable. Can anyone help this fledgling with his problems? FACTS... - Small US office has 56kb satellite link to India office - 56kb connected using Bay Networks AN Router & IP/RIP - Same US office just brought up T1 Frame-relay line through national ISP - T1 connected using Cisco 1005 Router & ....I think IP/RIP?? - US office server is WinNT 4.0 & workstations are Win95 PROBLEMS... (This I know of so far) - Office computers (PCs) can use either 56k or T1, but not both - If both are designated in PC's TCP/IP Protocol properties, can't get either to work - People in INDIA office cannot see T1 line or Internet at large GOALS... - How do I get Routers working together so that US PCs can see India & Internet networks and India PCs can see US & Internet networks? Sorry for my ignorance. I'm learning some stuff from InterNIC, Cisco & BayNetworks, but thought one of you might help accelerate my learning curve. Thanks, Erik _______________________________________________________________________ Erik Britt-Webb ~ VJ Infosystems erik(a)vjinfo.com Tel:(603)575-8546 ~ Fax:(630)575-8549 (12/96) www.vjinfo.com Total Commitment ~ Total Solutions ~ Total Satisfaction _______________________________________________________________________

3 2

BGP / Requirements, etc.
by Jamie 13 Nov '96

13 Nov '96

Hi, We're a "regional" ISP in Cleveland and Akron, OH, and are in the process of adding 2x10Mb links out. The new uplink wants to peer between us and our current backbone, running BGP. Our current backbone told me that it would be difficult to get full BGP running on our router: a Cisco 7010. RAM is not a problem, we can bump it up to 64M in a moment's notice. Has anyone had problems running full routes on 7010s? Caveats? Things to look out for? Also, one newbie-type question, our current backbone said that they "may not" pass traffic routed via backbone #2. If I am running full BGP between both backbones, what would the technical limitations be of passing traffic between all three sites (Backbone A, Me, and Backbone B)? Any BGP gods out there for hire on a consultant basis? :-) TIA, -jamie -- jamie g. k. rishaw | jamie(a)arpa.com | jamie(a)multiverse.com url-free sig file | the religious right | Multiverse - the internet company corp: 216.771.0002 | is *neither*. | Multiverse - your internet company C4 48 1B 26 18 7B 1F D9 BA C4 9C 7A B1 07 07 E8 .. on a sesame - seed bun

3 2

Re: BGP / Requirements, etc.
by Tony Li 13 Nov '96

13 Nov '96

Our current backbone told me that it would be difficult to get full BGP running on our router: a Cisco 7010. RAM is not a problem, we can bump it up to 64M in a moment's notice. Suggest you get your backbone to fire whomever told you that. From a BGP perspective, the 7010 and the 7000 are identical. It will be sufficient for a while (time TBD depending on net conditions). Also, one newbie-type question, our current backbone said that they "may not" pass traffic routed via backbone #2. If I am running full BGP between both backbones, what would the technical limitations be of passing traffic between all three sites (Backbone A, Me, and Backbone B)? You _could_ technically, provide transit between the two backbones in this situation. It sounds like your current backbone does not want that. Wholly understandable. You should configure your BGP to only advertise your routes to both backbones. Any BGP gods out there for hire on a consultant basis? :-) How much money do you have? ;-) Tony

2 1