Interplanetary Internet protocol in devel
shadowlight1 writes "This MaximumPCMag article discusses NASA's current research into interplanetary protocols for Internet data. The research team includes one of the inventor of TCP/IP. Get ready to ping-flood Pluto.
"
6 * * *
7 * * *
8 * * *
---
Bandwidth to Pluto isn't neccesarily going to be all that bad. Latency is what's going to suck. If it takes you 30 seconds to get a packet from point A to point B... that tells you nothing at all about the bandwidth. This is nonintuitive to some people. A common saying back when usenet was done via UUCP instead of TCP/IP was "it's hard to beat the bandwidth of a station wagon full of mag tapes". A station wagon full of media has a *tremendous* bandwidth, but really poor latency and a huge "packet size". UUCP is actually more suited to interplanetary communications than TCP/IP is. Luckily, we've got some great tools for getting UUCP networks and TCP/IP networks to play together nicely -- mail and news will work without a hitch over UUCP, even today. And MX records mean never having to say "I hate bang paths".
Sigh... we need more military people in /.
Space Communications Protocol, what the author must have been talking about, first of all, is being headed by MITRE (the strap-on brain of DoD back east to help them with anything geek).
The main goal is to delelop a protocol that looks and feels to the user like TCP/IP, but handles the fact that the major reason for packet loss being.. well, lost or damaged packets, literally, out into space.
TCP/IP assumes that lostpackets are because of network congestion, and so a missing packet is requested to be retransmitted.. and this usualy does the trick.. since most terra-nets run on fiber or copper...
If you kept asking for retransmissions in space - you exasserbate the problem so that if the errors grow to only 10^-6, and you use plain ol TCP/IP, the overhead and loss drowns the network out.. and you get nothing.
10^-6 errors can be a good day around here in the space biz... so one of the major points of SCPS is to deal with high BERs differently than TCP/IP, the other, of course, is security (how can you get spy sat data to the ground and beam it with an RF signal that anyone can pick up?)
SCPS has standard ftp, and will encorporate http eventually.. but its not done yet AFAIK.
You can read all about it here...
http://bongo.jpl.nasa.gov/scps
guns kill people like spoons make Rosie O'Donnell fat.
When it absolutely, positively, has to be lost at the speed of light interplanetary tcp/ip!
--