Very Computer

What are criterias for a TCP/IP stack on a embedded webserver?

There a many models on the market, but how can I evaluate them?

regards,
gerd

Hi Gerd,

it all depends on what exactly you want to do.

Whether your TCP/IP stack is only an add-on to an existing control
application of some kind or whether it is the purpose of the device itself
to BE the Server and server only.

There is one thing in common with all embedded servers (8-bit  and 16 bit
microcontrollers):

The RFC are not completely implemented for performance reasons - both in
runtime and program space.
Program space on microcontrollers is expensive, so this will have to be kept
at a minimum.

There are two or three TCP/IP stacks in the net that can be downloaded at no
cost:

Adam Dunkels : lwIP
Adam Dunkel's : uIP

violasystems : opentcp

and

Ethernut, which is mainly written for the Atmel AVR 8-bit RISC MCU, but can
be retargetted with some effort.

We offer a commercial grade product (which has to be licensed for money,
source license , no royalties) for which you will receive all sources and
lowlevel driver modules for the architecure you want to use.
So you can start working on your application right away without having to
spend days to get the adaption to your target implemented, tested and
released.

Also, as I can see from direct comparison of our product vs. the stacks
named above, - the performance in both execution speed AND code size is
better.
Well, -- you may interpret this as marketing blah blah, but if you are
interested in a comparison feel free to contact our office (see footer).
I will not publish the results in this newsgroup because this would not be
nice to Adam Dunkels and the lwIP programmers and the violasystems people.
They put a lot of effort into their software and it suffices for many
purposes but if you have a commercial grade stack there is something beyond
that makes it worth investing real money into that kind of software.

If you want to evaluate:
download the stack and implement a test application

Commercial software:
Contact the vendor (us :-) ) and ask for a demo and specs.

Hardware (TCI/IP on-chip) ask for datasheets and ask for a demo kit or buy
one.

It all depends on your budget.
-- If you are short on dineros but time is available : use a freeware stack.
-- If time to market is an issue: contact us. (or other vendors like CMX
(www.cmx.com) , Segger (www.segger.com))

Hope this helps you to get started

with kind regards

--
-----------------------------------------
Jan Homuth
Application Engineer
Technical Support , Embedded Tools

Altium - Making Electronics Design Easier

Altium Germany GmbH
Technologiepark Karlsruhe
Alber-Nestler Str. 7
D-76131 Karlsruhe
Phone: +49 721 8244 300
Fax:   +49 721 8244 320

WWW: http://www.altium.com
-----------------------------------------

Quote:> What are criterias for a TCP/IP stack on a embedded webserver?

> There a many models on the market, but how can I evaluate them?

> regards,
> gerd

Quote:> I will not publish the results in this newsgroup because this would not be
> nice to Adam Dunkels and the lwIP programmers and the violasystems people.
> They put a lot of effort into their software and it suffices for many

Huh. "I'd like to tell you how good our buyware product is but I don't
want to hurt the feelings of our freeware competitors". Egregious
self-righteousness of the worst kind. They may have told you to talk
this way in Marketing 301, but Marketing 301 is not aimed at
engineers.

(Speaking in general, since I am not affiliated with any of the
projects you referenced) - If you have specific benchmarks that show
some buyware product of yours is better than some free product of
mine, it doesn't hurt me if you publish those. There is no shame to me
in making a free product that underperforms a commercial product,
though I naturally reserve the right to improve any area I choose of
my free product in competition with yours.

Your smirking comment begs the questions:

Could it be that your benchmarks are artificial and meaningless, like
practically all other benchmarks ever devised, and that they won't
stand up to close expert scrutiny? Are they perhaps suitable for
publication only in private to people who have previously admitted
they don't know much about the technical details of this field?

Could it even be that you have no actual benchmarks?

Pfah.

Dear Mr. Edwards,

I just wanted to be polite. (You do not need to attend a marketing course
for that, which by the way I never did)

If you have any doubts about what I stated in the last message:

I know hat I am talking about, since

a) my everyday business is to support professional developers using our
tools EVERYDAY.
b) I am part of the development and testing process of the product I
described.
c) What I said about the freeware stacks available corresponds to my
personal experience on various architectures.
I did not mean to say it's crap.
-- to get this absolutely straight:
The stacks mentioned are very well done software.
Is that clear enough ?

d) I would not say something like that if I had not actually run the stuff
on real hardware.
e) it is my job to be polite and not an *
f) The question is so general that coming up with details is overkill at
this stage.

Now, please let Mr. Moser decide about the quality of the reply and cool
down a bit.

I did not mean to offend anyone.

Hope that settles the dust a bit.

With kind regards

Jan Homuth

Quote:> > I will not publish the results in this newsgroup because this would not
be
> > nice to Adam Dunkels and the lwIP programmers and the violasystems
people.
> > They put a lot of effort into their software and it suffices for many

> Huh. "I'd like to tell you how good our buyware product is but I don't
> want to hurt the feelings of our freeware competitors". Egregious
> self-righteousness of the worst kind. They may have told you to talk
> this way in Marketing 301, but Marketing 301 is not aimed at
> engineers.

> (Speaking in general, since I am not affiliated with any of the
> projects you referenced) - If you have specific benchmarks that show
> some buyware product of yours is better than some free product of
> mine, it doesn't hurt me if you publish those. There is no shame to me
> in making a free product that underperforms a commercial product,
> though I naturally reserve the right to improve any area I choose of
> my free product in competition with yours.

> Your smirking comment begs the questions:

> Could it be that your benchmarks are artificial and meaningless, like
> practically all other benchmarks ever devised, and that they won't
> stand up to close expert scrutiny? Are they perhaps suitable for
> publication only in private to people who have previously admitted
> they don't know much about the technical details of this field?

> Could it even be that you have no actual benchmarks?

> Pfah.