We are coming out with new products - ala PM-3. Reengineering existing
products is pointless. They work, and they still have growth potential.
You just have to carefully consider what you add.
>shcool department almost buying a box from Cayman that now seems like it
>must have been a PM2e10 in disguise...and that was a few years back.
It was. We built them.
>A 30 port terminal server / router with 1mb RAM sounds a little
>prehistoric to me. First thing I did when we got a PM2 was upgrade it to
You don't understand then. The PM can do almost everything it is used for
with 1Meg. It is only say in the last 6 months that the memory usage
has grown so that some users find it worhtwhile to upgrade to 4Megs. But
the vast, overwhelming majority of PMs running are running fine with 1Meg.
We run ours here in default. Engineering is planning on recompacting the
code again to make more room in flash. I believe 3.3 was smaller than 3.1.4,
but we've added a LOT of features from 3.3 to 3.5. Time to trim it back.
Look at the ComOS, and then look at the oper image for an Annex or the IOS.
ComOS 3.3.3 is 357K, 3.5b17 is 389K currently (betas tend to not be fully
optimized). Compare that to the multiple megabytes some other comm servers
OS's are. Much of the advantage comes because the ComOS was created from
scratch, to run comm servers, and it hasn't been laden with every bell and
whistle peoplr ask for. New features are carefully considered - which is
a major reason for the overall reliability of the ComOS. We definitely have
bugs - but if you watch competitors lists, USEnet, etc - you see that we
tend to have far fewer problems.
Look at the HW. The CPU in the PM-2 still has overhead to grow. People
keep claiming it must be out of horsepower - they are dead wrong. Other
systems will max out before the CPU (barring spurious interupts - we're
talking normal operation) - and no, they still have room too. As I have
said countless times - directly comparing CPUs is *meaningless*. It is just
one factor in many - software efficiency and overall architecture are more
important.
A PM-2e-30 has 30 115.2K serial ports with one 386 processor and 1Meg RAM.
I compare it to the Xylogics RA4000 with 36 ports (why? Because I used to
work there and I know that better than another competitor. Nothing against
Xylo). That unit has 36 115.2K serial ports with 2 486 processors and a
minimum of 4.5Megs of RAM.
Performance of the two units is comparable, each has a part of the curve
where it has a slight lead. The RA is much more expensive - AND by default
it cannot self boot, that is an option. It needs a boot host. I believe
the OS is around 1.5-2Megs. It's been a while since I saw a current image
and they've split the Annex and RA trees so that probably impacted size.
This is just a direct example of why comparing processors and judging
performace by chips is a pointless activity. By that method you'd look
and see 1 386 vs 2 486s and would probably expect the 2 processor unit to
far outperform the 1 procssor unit. Which just isn't the case.
Same thing with some of this units that brag about using high end RISC
processors, yet they're performance isn't stellar. Why? Much of the time
it is because they don't do distributed processing - that CPU is working
hard to do everything. Livingston units distribute the processing tasks,
the PM-3 is the best example of this.
There are abilities in the PM-2, and other units, that have yet to be
tapped.
-MZ
-- Livingston Enterprises - Chair, Department of Interstitial Affairs Phone: 800-458-9966 510-426-0770 FAX: 510-426-8951 megazone@livingston.com For support requests: support@livingston.com <http://www.livingston.com/> Snail mail: 4464 Willow Road, Pleasanton, CA 94588