At 08:53 AM 5/15/99 -0400, you wrote:
>
>Denis,
>
>Ingress processing for Linux now is available for Linux. Still being tested.
>Measurement tools etc are also in the works. Your noise contribution
>on this list is going to be very limited, Soon.
LOL..now available..still being tested. Just like bridiging, x.25 and lots
of other things that dont work. You are smudging the os by announcing that
things are available that are unusable by anyone with real needs. That is
one of my points. Every time someone gets something "kind-of working" its
becomes a feature of the OS.
>
>Now in regards to BSD: If Linux was so useles, why are you
>bothering with drivers for Linux?
Its not useless. What is wrong with you dopey people? Its just a lot more
work to do the same things and more likely to have things that dont work
well. Our linux drivers work very well. There are lots of people running
linux system s than need T1 and T3 cards. It can be made to work...but if I
was starting from scratch and didnt have experience with either one, I'd
chose BSD. People who switch to BSD rarely go back. Its just a cleaner O/S.
>You might refer to the consistency in BSD as stability; some of us
>look at it as lack of progress. You are still stuck with mbuffs,
>for example.
Yes I do. People with T3 lines can't afford to wait 6 months for gateD to
stop being flakey.
You think skbuffs are an improvement? Lets attach every piece of info we
can to the buffer and pass it around. And when are they gonna add chaining
to sk_bufs so you dont have to allocate enormous receive buffers for large
MTUs? Pretty efficient to allocate a 5000 byte buffer because you might get
an occasional large frame on a serial line. mbufs are only slower when
chaining is used, and now chaining is only used when necessary. skbufs are
an amateurish concept and break the rules of layered networking. sk_bufs
were the major cause of instability in 2.0 kernels.
>
>For the interested: The June issue of the Linux Journal has an article
>on IP Bandwidth Management (that I wrote). It might shed some light
>on what is available on 2.2
>Also, the Ottawa Linux Symposium will feature a tutorial on how to
>use these features. A live setup (and hopefully some of the tools freely
>available then) will be part of the demo. Sign up at: linux.achilles.net
What is available is irrelevent. You have to be a linux hacker to get it to
work, and thats the point. People dont want to dick around with the os for
days to get it to work. At some point, perhaps the stuff in linux will be
as effective as our ET/BWMGR, but it will never be as easy to use or
install or as useful to a specific target group as ours is.
>
>Thanks for providing the inspiration, Denis. Maybe this will inspire you
>to shift your strategies.
LOL...selling to hackers like yourself has never been a "strategy"...guys
who would rather spend hundreds of hours struggling with experimental code
rather than pay $495. for a proven commercial product are nobody's "target
market". When you've got 3500 hosts limited in production environment then
you can talk. And by the time you do, I'll be 10 steps ahead of you.
There's lots of "bandwidth manager" products that dont work very
well...writing articles and producing commercial quality products are two
different animals, my friend.
Dennis
Note the 2 "n"s jammmal. :-)
Emerging Technologies, Inc.
http://www.etinc.com
ISA and PCI T1/T3/V35/HSSI Cards for FreeBSD and LINUX
HSSI/T3 UNIX-based Routers
Bandwidth Manager
http://www.etinc.com/bwmgr.htm
-
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to [EMAIL PROTECTED]
