Statistics please. Including the firmware used by each.
> > I can tell you with certainty that there are crappy K56Flex modems.
> There
> > are also really *good* ones. Zoom happens to make a good unit (gasp!)
> and
> > so does TDK in the PCMCIA arena. I have a TDK CyberExpress, and it is
> > hands-down the most stable and best-connecting PCM card-modem I've found.
>
> Rockwell seems fond of the Zoom as well.
That's because it works.
> We used Hayes Accura, Motorolo
> Modemsurfr, Diamond Supra, and Zoom. All were confirmed as having the
> latest firmware.
Oh hoh! Well, there's part of your problem right there. Zoom's 1.2
firmware bites. Of course, if you talked to the tech departments at any of
these ISPs, you would have found that out and back-flashed to 1.09 or 1.02.
3COM/USR has never put out a bad firmware load, have they? (cough cough!)
> I never could get much difference
> statistically or operationally between the various modems with Rockwell
> chipsets. They all worked about the same and your theme that some are
> better than others is not detectable as far as I can tell. They were
> remarkably uniform.
Then your testing was flawed in some other way, because that just simply
isn't true.
> > This doesn't matter for X2, since there is only one vendor (3COM) making
> > said central-site hardware, and all code is licensed from 3COM on the
> > client side as well. Thus, there really is only ONE implementation.
> >
> > But this is NOT true for K56Flex. To indict a protocol in a multi-vendor
> > environment when it is KNOWN that the huge majority of the hardware you
> > are dialing into is made by one vendor is to indict the wrong thing!
>
> Well, you're explaining/apologizing for WHY it happened. But you're hardly
> denying that it did. Your "one vendor" central site stuff is simply in
> error Karl. Get off of it. You don't know what you're talking about, and
> it simply isn't true.
Prove that its statistically invalid Rickard. Post the statistics. How
many K56Flex ports did you attempt to connect to, and how many of them were
made by which vendor(s)? What *firmware revisions* were involved? "Latest"
means nothing without *NUMBERS*.
If you were running 1.2 in the Zoom modems, or its equivalent in others,
well, that explains a lot. But instead of trying to find out WHY, and take
care of the problem (gee, again, nobody has ever produced a bad firmware
load, right?) you just slam an entire technology.
> > IF you want to claim that *ALL* K56Flex implementations suck (which is
> what
> > Rickard is doing), then you have to document that ALL K56Flex hardware
> and
> > the connection points involved sucked - and the existance of just ONE ISP
>
> > running a given piece of hardware which doesn't suck proves the converse.
> >
> > Averages in this kind of environment, as were quoted, are MEANINGLESS.
> >
>
> Averages are always quite meaningful Karl. That's why people do them.
They're only meaningful if the statistics measure what you're claiming they
do. You're not. If you were, then NOBODY would be getting the good
connections you claim are impossible.
> I
> can't imagine what point you're making here. If you can do excellent
> K56flex in a limited area, and deplorable connections everywhere else, how
> does this prove the converse of anything?
A limited area?
> I would think it would be
> desireable to have the best performance over the widest possible variation
> of network topography. I can't believe they would release it if it didn't
> work SOMEWHERE. THe point is, over a very wide footprint, the two do not
> produce not peer results.
You weren't testing what you claim to have been testing.
> > Now, Rickard hasn't disclosed the details. He knows that to not do so is
> > going to mislead - or, if he DOESN'T know, he's not qualified to do this
> > kind of evaluation. Yet he's doing it anyway, and failing to disclose
> > material facts which bear on the conclusions.
> >
>
> Now what would you know about what I know, and what we are going to
> disclose? These wild conjectures are not even rational.
So why not tell us what you are going to disclose Jack? Or is it too late
to hastily re-write the commentary and reprint, so you'll obfuscate instead?
> > The bottom line from my point of view is this - we run a whole lot of PM3
>
> > hardware. We have the connection rate statistics and termination reasons
>
> > to *prove* that it doesn't suck.
>
> Well, if you're just really smacky darn happy with it Karl, go and be well
> and enjoy.
If we weren't, Livingston would have gotten them back.
> > Those people who can't get K56Flex (or any PCM) connections *AT ALL* are
> > going to get connections around 30kbps (either 28.8kbps or 31200, most
> > likely). But those who *do* get K56Flex here typically are receiving in
> > the mid 40s - exactly what Rickard is claiming X2 delivers, and K56Flex
> > does not.
>
> Well, that was the icky part actually. We DID get K56flex connections. If
> we just hadn't gotten any, it would all make sense. The highest speed call
> recorded in the test was a 54000 connection with K56flex. It was never
> repeated. It occurred once in 145000. But we had hundreds of calls in the
> forties. They were just grossly outnumbered.
Your test was biased.
> > I'll be happy to compile and post the stats if I have to to prove it.
>
> Sure. Do your own test. Start your own magazine. You go girlfriend.
Oh boy, more gratuituous personal attacks. Not that I'm surprised.
> > I just "spot-checked" a half-dozen servers. There was not ONE PCM
> > connection running under 40kbps. NOT ONE. There were a few 50kbps
> > connections, and most were either 44kbps or 46kbps (of the connections
> > running in PCM mode). There was one (ONE!) 40kbps connection - that was
> > the worst PCM-mode link online of the random sample I just took.
>
> Well, I guess your scientifically exhaustive study pretty much wipes out my
> silly 145,000 calls. I mean you DID go look and everything and it SEEMS
> hunky dorey back in the old equipment room. I give up. You're right. It's
> a peachy system. Didn't mean to disparage it Karl. Order more. Go and be
> well.
Your silly "145,000" calls (if they were really made) is really rather small
compared to what our infrastructure records.
Nothing "seems" ok Jack. We have the hard statistics and connection records
to prove our claims, and we're not afraid to disclose them.
Livingston/Lucent is welcome to my connection data for the purpose of suing
you off the planet if they want it. I only have a couple of years of data
across more than 10,000 users and tens of thousands of sessions a day.
I'll post a summary - it ought to be interesting to see you try to argue
with the facts.
---- Karl Denninger (karl@MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin http://www.mcs.net/ | T1's from $600 monthly to FULL DS-3 Service | NEW! K56Flex support on ALL modems Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS Fax: [+1 312 803-4929] | *SPAMBLOCK* Technology now included at no cost - To unsubscribe, email 'majordomo@livingston.com' with 'unsubscribe portmaster-users' in the body of the message. Searchable list archive: <URL:http://www.livingston.com/Tech/archive/>