- Test - lists.nanog.org

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

spamspamspam (fwd)
by dgaudet＠plebe.com 13 Nov '96

13 Nov '96

And yet again the usefulness of whois contacts is eroded. I don't know about the rest of you, but I'm beginning to indirect all my whois information through idiot filters -- indirect email addresses, alternate phone numbers. I don't even run anything large. What's going to be left to contact other network operators if whois becomes useless? Hopefully the RADB... if everyone were in it. At least this program below doesn't use prtraceroute to find upstream providers, yet. Dean ------- start of forwarded message ------- Date: Wed, 6 Nov 1996 23:54:50 GMT From: steveha(a)net-services.compulink.co.uk (Steve Harris) Get lots of e-mail offering you get-rich-quick schemes? Want to hit back? "Spam Hater" is free Windows software that helps you respond effectively and makes it hot for these people. * Analyses the Spam * Extracts a list of addresses of relevant Postmasters, etc. * Generates a "WHOIS" query to help track the perpetrator * Prepares a reply * Choice of legal threats, insults or your own message * Appends a copy of the Spam if required * Puts it in a mail window ready for sending Spam Hater works with lots of popular e-mail programs directly - there's no tedious cutting and pasting. Supported E-Mail Programs Ameol, AOL 2.5I, Eudora Light 1.5.2, 1.5.4, Eudora Pro 2.2, Free Agent 0.99, Microsoft Internet Explorer 3.0 (4.70.1155), Netscape 1.2N, 2.02 3.0 Pegasus Mail V2.4X, Virtual Access V3.51. Download from http://www.compulink.co.uk/~net-services/spam/ ------- end of forwarded message -------

5 7

Re: Why doesn't BGP...
by Vadim Antonov 12 Nov '96

12 Nov '96

I'm actually quite sure -- the reaction time of the control unit should be less than the characteristic period of the controlled system. Think of it as a big pendulum which is subjected to rapid kicks -- by applying kicks at different directions you can keep pendulum at any angle. Try that if the pendulum swings faster than you can kick :) Not that it is very useful for updating metrics, but you can imagine a car radio which makes congestion announcements fast enough, so as to load 101 and 280 equally (car here, car there). Wait a second, that reminds me of something :) --vadim From: alex(a)relcom.eu.net Subject: Re: Why doesn't BGP... Vadim, are you shure about this? Withouth the friction - what would you got this case? Very big pendulum, isn't it?

1 0

Re: Why doesn't BGP...
by Vadim Antonov 12 Nov '96

12 Nov '96

Peter Lothberg <roll(a)Stupi.SE> wrote: >On the radio1:: > "101 north is congested, take 280 instead" > >Guess what hapens 5 minutes laster; >On the radio2:: > "280 north is congested, take 101 instead" >goto on the radio1 Goto the theory of control systems :) This is a textbook case of a system with too strong negative feedback loop. There are two ways to deal with that -- reduce the feedback (say, make the announcement in a barely intelligible speech, so only 1/3rd of drivers will get it :), or increase inertia of the system (by posting 10MPH speed limit :), so announcements can come fast enough to compensate each other. Seriously, the load-sensitive traffic management _can_ be done, but to do it you a) need full knowledge of the load patterns, plus some historical data, and b) have a way to control system in steps less drastic than rerouting an entire flow. Can't be done with ciscos, or any existing routers... --vadim

5 4

RE: Why doesn't BGP... Why not? I'm not redistributing.
by Chris A. Icide 12 Nov '96

12 Nov '96

Ed, The problem being, you would need state information from the entire Internet (or at least to the networks that you are trying to reach), forcing everyone in the path to any of those networks to run the same protocol and maintain state information. Currently, the technology to perform such a function is available only in pre-production equipment. Most of the core routers of the internet are already struggling with just maintaining the packet flow. The addition of a new global state protocol would bury the things. Chris ---------- Ed's comments appended I'm basically a stub network. I take routes from 3 AS's right now, but I don't redistribute anything I learn. The only redistribution that goes on for me is in my iBGP peering. The convincing argument for me against doing the "intelligent" route selection has been related to the huge route flapping that would ensue. My question is, if I'm not redistributing the routes (except internally, and I have a certain tolerance for route flapping internally) then why not do the intelligent route selection? (the route selection would of course have dampening parameters etc.. and be turned off by default, but have the ability to be enabled, etc) Ed Henigin ed(a)texas.net

1 0

Re: Why doesn't BGP... -Reply
by Neal Castagnoli 12 Nov '96

12 Nov '96

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. 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? --Neal >>> Avi Freedman <freedman(a)netaxs.com> 11/09/96 05:31am >>> I understand that IOS 12.0 will solve the travelling salesman problem for arbitrary number of cities. Seems they had to find a way to solve NP-complete problems in sub-exponential time in order to give us better algorithms for guaranteed-optimal routing. Anyway, sorry to be snide - any suggestions you have about algorithms for exchanging routing data between (or inside of) providers that also takes into account link size and perhaps hop-count, as opposed to just AS-PATH, are welcome at the next IETF meeting. Avi

1 0