January 1997 - Test - lists.nanog.org

Re: Film at 11:00
by Vadim Antonov 02 Jan '97

02 Jan '97

> Well, i do not think that using off-the-shelf cheapo CPUs debugged by > somebody else and tested in millions of applications, instead of > single-purpose ASICs qualifies as a reliability hit or a development lag :) >Agreed that the part itself is correct. However, unlike you, the rest of >us tend to take perfect parts and make mistakes putting them together. ;-) Ah, i do a lot of mistakes. Anyway, it's much easier to put together debugged parts, than put together new silicon. I've had more than enough of that at nCUBE. --vadim

1 0

Re: Film at 11:00
by Vadim Antonov 02 Jan '97

02 Jan '97

Tony wrote: > Providing that there's a way to get a bunch of cheapo CPUs to do the > job. >True, but there's a complexity hit (which translates into a reliability hit >and a development lag) from this approach. Well, i do not think that using off-the-shelf cheapo CPUs debugged by somebody else and tested in millions of applications, instead of single-purpose ASICs qualifies as a reliability hit or a development lag :) --vadim

2 1

Re: Film at 11:00
by Mike Vallee 02 Jan '97

02 Jan '97

At 10:35 AM 1/3/97 -0500, Avi Freedman wrote: > >Well, it's also somewhat more complicated to distribute the tasks in a router >to multiple CPUs. > >There are plenty of ways to break Bays by accidentally doing that distribution >wrong. > > >Avi > Hmmm. There are some who would contend that there are plenty of ways to break Bays by accidentally doing something right. ;) mike v

1 0

Re: Which RFC?
by Paul G. Donner 02 Jan '97

02 Jan '97

Alex: While you are probably looking for "mandate" material, OSPF provides a specific example of implementation of the most specific route, in an actual routing protocol. Check out the OSPF RFC (I believe it was 1247) as well as the new one (1583). The old one talks about selecting the least ambiguous route. The new RFC I believe has changed the terminology to use "best match" or "most specific match" as opposed to "least ambiguous". >From Section 3.5 IP subnetting support (RFC1583) "When an IP packet is forwarded, it is always forwarded to the network that is the best match for the packet's destination. Here best match is synonymous with the longest or most specific match." >From Section 1.1 Protocol Overview (RFC1583) <bigger>"OSPF calculates separate routes for each Type of Service (TOS). When several equal-cost routes to a destination exist, traffic is distributed equally among them. The cost of a route is described by a single dimensionless metric.</bigger>" In OSPF's case there doesn't seem to be any implication of metric involvement in forwarding from the RFC. The metric is used to define the shortest path for routes to be implemented in the topological database. OSPF works well with subnetting. Each subnet carries its mask with it and these are also both present in the routing table. Donner At 03:16 PM 1/2/97 +0000, Alex.Bligh wrote: >Anyone know *which* RFC says a packets should be routed using the >most specific route in a routing table, not (for instance) the first >route in the table that matches, or, for instance, using a less specific >route that has a better metric? This is so basic I hardly >know where to find it, but I have a (for the time being anonymous) >systems vendor who really prefers the way their kernel does it .... >Oddly enough it doesn't work well with subnetting. > >Alex Bligh >Xara Networks > > > >

1 0

Re: Which RFC?
by Kamau Wanguhu 02 Jan '97

02 Jan '97

RFC 1338 makes mention of this in regard to CIDR. Later Kamau Wanguhu ---------- > From: Alex.Bligh <amb(a)xara.net> > To: nanog(a)merit.edu > Cc: amb(a)xara.net > Subject: Which RFC? > Date: Thursday, January 02, 1997 10:16 AM > > Anyone know *which* RFC says a packets should be routed using the > most specific route in a routing table, not (for instance) the first > route in the table that matches, or, for instance, using a less specific > route that has a better metric? This is so basic I hardly > know where to find it, but I have a (for the time being anonymous) > systems vendor who really prefers the way their kernel does it .... > Oddly enough it doesn't work well with subnetting. > > Alex Bligh > Xara Networks > >

1 0

Re: Which RFC?
by Bill Woodcock 02 Jan '97

02 Jan '97

> Anyone know *which* RFC says a packets should be routed using > the most specific route in a routing table, not (for instance) > the first route in the table that matches, or, for instance, > using a less specific route that has a better metric? It might be worth your while to look through the sections of RFC 791 and subsequent revisions which define the second byte of the IP header, which sets the delivery-priority and the class-of-service for the packet. The class-of-service (type-of-service?) field lets you specify whether you want the shortest, largest, cheapest, or most reliable route, those being mutually exclusive. In my experience, both fields are set to 0 on just about all packets, which doesn't specify any requested class-of-service. That presumably leaves the router free to choose any route it likes, at its discretion. > This is so basic I hardly know where to find it If the router, in its discretion, chooses to route the packet in the wrong direction, that should be easily demonstrated and reasonably incontrovertable, no? What's your vendor's argument in favor of shipping the packet to a nearby site which isn't associated with your destination? I'm kind of curious to hear it... Who's the vendor? Please humiliate them publicly. :-) -Bill ______________________________________________________________________________ bill woodcock woody(a)zocalo.net woody(a)nowhere.loopback.edu user(a)host.domain.com

3 2

Which RFC?
by Alex.Bligh 02 Jan '97

02 Jan '97

Anyone know *which* RFC says a packets should be routed using the most specific route in a routing table, not (for instance) the first route in the table that matches, or, for instance, using a less specific route that has a better metric? This is so basic I hardly know where to find it, but I have a (for the time being anonymous) systems vendor who really prefers the way their kernel does it .... Oddly enough it doesn't work well with subnetting. Alex Bligh Xara Networks

3 2

Re: Film at 11:00
by Vadim Antonov 02 Jan '97

02 Jan '97

Tony Li wrote: >So the problem is that as we approach higher speed (post-7500) interfaces, >you're forced into more specialized hardware. At this point, not having an >SSE ASIC becomes somewhat silly. The rest of the board is ASICs, which in >volume end up cheaper than processors... There's not enough volume in high-end box market to get to the break-even point on the price-performance. Providing that there's a way to get a bunch of cheapo CPUs to do the job. And don't forget the DRAM vs SRAM issue. I did the arithmetic. --vadim

2 1

Film at 11:00
by Paul Ferguson 01 Jan '97

01 Jan '97

Looks like the 45k mark was reached: Table History Date Prefixes 231296 43352 241296 44089 251296 43735 261296 43391 271296 43093 281296 43912 291296 45352 301296 45532 Happy New Year on the Internet, - paul ref: http://www.employees.org:80/~tbates/cidr-report.html

5 4