[LINK] RFI: Packets per Message?

[Stilgherrian]

On 22/4/05 1:57 PM, "Roger Clarke" <Roger.Clarke at xamax.com.au> wrote:
> "Generally, messages that are meaningful to users, or to the
> application software that they use, are broken into multiple packets
> for transmission over the Internet"?

I think this implies that the decision to break data into packets is based
on the meaningfulness of the data content to humans -- perhaps avoid this.


> (I'm reviewing a paper to appear in IEEE Spectrum in June, i.e.
> deadline is upon them.  The readership is educated professionals -
> think power-station engineers and suchlike, rather than IT types).

If they're not IT folks, do you need to get into a discussion of what kinds
of data are broken up into packets? That seems a bit geeky to me. Or are you
trying to emphasize that anything apart from "routine technical messages" is
turned into a stream of datagrams?

Even a short message is still a datagram. And even an HTTP request of only a
few bytes will still need multiple packets to set up the data stream.

So maybe this is the go:

"All data transmitted over the Internet is sent as small packets called
'datagrams' -- typically less than 1500 bytes in size, although the exact
size varies depending on the standards of the underlying physical network.
Simple messages might consist only of one datagram. A typical example is the
exchange of routine network information when a domain name is translated
into a numerical Internet address. But most data which humans generate --
web pages, software downloads, images, email, MP3 files and streaming
multimedia -- will consist of a number of packets, maybe even thousands of
packets in the case of large files."

HTH,

Stil


-- 
Stilgherrian <stil at stilgherrian.com>
Internet, IT and Media Consulting, Sydney, Australia. ABN 25 231 641 421
mobile 0407 623 600 (international +61 407 623 600)
fax 02 9516 5630 (international +61 2 9516 5630)