On 7/31/08, Steve Jones <[EMAIL PROTECTED]> wrote: > 2008/7/31 Mark Baker <[EMAIL PROTECTED]>: > > The world is a very large place, and includes a lot of companies that > > can potentially do business together. Any automated solution to > > facilitate those business transactions will necessarily produce enough > > Internet traffic to show up on graphs like this one. > > > > http://www.caida.org/data/realtime/passive/?monitor=sdnap > > > Errr obviously I'm as thick as a whale omelet but tunnelling CORBA > over HTTP would appear as HTTP on that graph and SOAP over HTTP would > appear on that graph under HTTP....
I didn't mean that what would show up on that graph would be a new protocol, only that the traffic must be measurable at that scale. I suppose had I found a graph of traffic on a major backbone, that this might have been more obvious, but I couldn't in the time I allotted myself. My apologies. > Then of course there is a big chunk (2nd biggest chunk in fact) which > is unknown TCP and TIMBUKTU beats in HTTPS. Hell people could be > doing a massive chunk of SOAP over MQSeries for all we know in the TCP > block. There's dozens more options I'd pick ahead of those, including VPN traffic. > > If your solution doesn't appear on that graph, it simply doesn't > > matter to the world at large. > > > I'll phone someone at Vodafone and let them know that their entire > business fails to appear on an important graph... I'm really not sure that you appreciate the size of the network of inter-business messaging we're talking about here, Steve. Vodafone is insignificant at this scale, as is every other company you'd care to name. The traffic generated by large companies would obviously be considerably more, in general, than that generated by smaller companies, but any view of traffic focused on individual nodes is going to be insignificant compared to N-way traffic. > > They're pretty much all using HTTP and either XML or HTML URL-encoded > > forms (and some Javascript where UI integration was required), so > > they're already on the graph. > > > So are the HTTP/SOAP folks and the CORBA tunnelled over HTTP or the > VPN tunnelled over HTTP. All of those people are "using" HTTP but > "using" a protocol says nothing about "how" it is used. Move out of > the network layer and into the software. Sure, you're right, that does count as HTTP traffic in those graphs. Are you suggesting that if we removed IIOP/HTTP and SOAP/HTTP data from that graph, that it would look any different? Mark.
