Slashdot Mirror
Web Enabled Spacecraft
gilgsn writes "Yahoo has an article from space.com about a satellite which will be operated by FTP over TCP/IP on the Internet! The CHIP (Cosmic Hot Interstellar Plasma Spectrometer) spacecraft will examine the stuff between stars, the so-called void of space that is actually rich with hot gas. The choice of protocol was dictated by economics. I wonder what OS it will run and if communications will be encrypted?"
Though they may have anticpated the slashdot effect.
Watch out: the RIAA will shut down the first Mp3 pirate server in space!
Windows SPACE! Service Pack 12.
In other news, a satellite was taken over by a 5 year old...
help me i've cloned myself and can't remember which one I am
We were asked to come up with a fast cheap solution for getting two servers to keep in touch, and not have to change the company's firewall setup. Our solution was to use SOAP and JAXM, but our backup plan was to do everything via FTP...who knew we were in the same league as NASA?
I'm a writer, a poet, a genius, I know it. I don't buy software, I grow it.
Maybe it's just me, but doesn't it seem strange that it would be operated by FTP? I mean, one would think that being able to SSH to it and having a command prompt would be a lot more useful...
:)
Unless it's actually a cover and NASA is creating the first orbiting pr0n server
It wont take long for someone to crack the satellite. They will download images thinking to collect nice data about deep space but instead will find that some h@X0R has redirected their satellite to take a good close look at Natalie Portman.
Or...the first DDoS initiated from space will soon be in the headlines.
FTP? C'mon!
In Bushworld, they struggle to keep church and state separate in Iraq as they increasingly merge the two in America.
Wow, I've never heard of FTP being used as a control protocol. Sure, HTTP might have been a bit much (although I doubt it. people have run webservers computers the size of matchheads. Even HTTP over a serial connection on an apple2). Why not use straight telnet with no options?
And I wonder how this control works, do you CD into a spesifc cordinate of space to examine? Can you DIR the stars it can see to find which ones to look more closely at, and then GET the acual data?
Hrm, actualy that would be kind of cool.
autopr0n is like, down and stuff.
Couldn't they have picked better protocols? It seems to be me for reliability and performance that isn't the best of choices. There are alot of other protocols (XTP for example) that the government could have used instead. Although TCP/IP is so commonplace I wouldn't want my 15 million dollar satellite to depend on it.
What? No security through obscurity? I can hear the 5kr1p7 k1dd13z rejoicing...
J. |_337 H4x0r: D00d$$$ I took control of da satelite man! See hoe |337 I am!!!
J. |_3373r H4x0r: L00se, sux0r! I can make her spin round! Wheee! Wheeee!
J. |_337 H4x0r: What are you doing idiot you're taking her down!!!
J. |_3373r H4x0r: No way man. I'm much to |337 for that!! DAMMIT Windows crashed again! sux0rzzz!!!###
[Sattelite falls down to Earth]
Please correct me if I got my facts wrong.
Ok, satellite running windows, yada yada, service pack yada yada, hacked by 12 year old, yada yada, Microsoft Windows for Space yada yada.
Ok, now I've gotten all the blindingly predictable jokes out of the way, can we move to something more interesting?
Read reviews of shopping cart software
Hope it has an IP address, It would be cool to ping something not on earth.
In the first week after it goes online some kid will manage get a root shell, install Apache and leave a page stating that "IN SOVIET RUSSIA, DEEP SPACE PROBES YOU"...
Serious kudos to these guys for the work they're doing! From what it sounds like, they're using FTP inside of either a IPSec or custom layer 2 encrypted tunnel -- once you've been wrapped by that, you're mostly OK (though FTP servers in general have had some pretty nasty growing pains).
:-)
Some may be wondering why the use of FTP, instead of HTTP. Indeed, HTTP is a unified protocol capable of elegantly handling both (moderately) interactive command exchange and bulk data transfer. The problem is latency -- if this beast is going anywhere, there's going to be some significant (5-10 second, minimum) lag between issuing commands and receiving responses. In such an environment, you don't *want* interactive access; you want an elegant way of providing a series of commands and receiving a series of responses. FTP provides that -- among other things, while HTTP's capacity for downloading files is quite mature, anything more is asking a bit more of HTTP than it was designed.
FTP has specific commands for machine interaction w/ the file server -- NLST provides a standard formatted directory of files, independent of the underlying implementation. By contrast, Apache dumps some HTML.
WebDAV ("Web Folders) was meant to address complex file system operations under the rubrick of HTTP. Thus far, it hasn't been much of a success. It most likely never will be. Thus, FTP is used.
But FTP is built on TCP, and this introduces a problem: The affects of latency upon the underlying TCP error handling protocol. TCP implementations are notoriously untuned for the case of high bandwidth, high latency. They're built to assume the lack of a response implies either congestion on the line or packets being dropped; either way, implementations tend to scale back. Significant work has been done to address this case, mostly on the behalf of Satellite systems (the ultimate in high latency, high bandwidth access). Mostly, the idea is to expand window size (the amount of data that each side is allowed to send before it must receive an acknowledgement) to match the amount of data that's literally hanging amidst space and time on its way to its receiver. But this is a very hard problem, one of the few that the architecture of TCP has quite a bit of trouble scaling to handle.
NASA went to a bulk transfer protocol, partially because interactive performance across large distances is problematic. But the bulk transfer protocol itself is based upon an interactive error management protocol. It'd be interesting to repurpose an established protocol for error-handled bulk transfers for just this use...I'm certain one of the "reliable multicast" architectures out there would be an astonishingly elegant solution.
That's not to say they made the wrong choice with FTP -- particularly if they tuned their stacks well, and encapsulated themselves amidst lower layer security, great job! Just that there's lots of work in this arena left to do.
If I remember right, Vint Cerf and a couple of his colleagues were working on IP protocols suited for communication between Earth and Mars. We're talking *minutes* of latency! Now that'll be a hell of a hack
Yours Truly,
Dan Kaminsky
DoxPara Research
http://www.doxpara.com
Conceivably, you could even control the satellite by ssh'ing into it and running various command-line apps. If you wanted to be really cute, run a web server on the satellite and make it controllable with web forms... but that strikes me as just a little over-elaborate :)
For security purposes, they mention using "standard commercial applications" to encrypt the link. Presumably that means they're running a VPN of some description. As an additional security measure, you'd presumably want to hide the thing behind a firewall and give it a non-global IP address (somewhere in the 192.168.*.*'s, presumably) so that it simply can't be reached from the wider internet, and then (if it was *really* necessary) set the firewall up so that the appropriate people can tunnel through.
Actually, it would be interesting if we could get a /. interview with one of the people behind this satellite (and grill them about their security measures). Roblimo, are you listening?
Any sufficiently advanced technology is indistinguishable from a rigged demo
--Andy Finkel (J. Klass?)
Personally, I find it very intriguing that they've chosen FTP as the protocol, though it does make a lot of sense. Most of what the sattelite is intended to do will be done in a pre-determined manner. Very little will be done in real-time. As a result, most instructions will be able to be scripted, and FTP is an excellent way of uploading scripted instructions to the sattelite. TFTP would've been even better, had it not been for the lack of access controls.
Now, that much said, when do you think we'll see the first DDoS of zombie spaceprobes? =)
And, this isn't the first satellite to use TCP/IP, by the way. TCP/IP has been run over satellite links numerous times, most often to demonstrate TCP's shortcomings in relation to better methods.
note: that's not to say that TCP/IP isn't a fine protocol -- it's a perfectly reasonable way to do things on a low BER, low latency network (i.e., the majority of networks we commonly use). I'd have the same criticisms of someone trying to run, for example, SCPS on a terrestrial network. It's the wrong tool for the wrong job.
SSH is the protocol they kept using to open and close the doors on the moon base. :-D
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
They'll need the 30+ mile long poles to hit Ctrl+Alt+Del...
On the Unix side -
That "No route to host" error becomes more meaningful.
"Uptime" will relate to orbit, not system.
--- I wish I could hear the soundtrack to my life. That way I'd know when to duck.
OTOH, doesn't TCP use alot of SYN, ACK etc. to establish/close a connection? This could be a problem because it multiplies the round trip time which could have been avoided by using a special purpose protocol.
I'm also wondering if there would be a high error rate because of atmospheric disturbances and such. If so, TCP would be really useful because you get error correction for free.
The Ground Systems Department at NASA Marshall Space Flight Center has a "new" system called Telescience Resource Kit (TReK) that allows experimenters to hook personal computers in their home labs up to experiments they are running aboard the International Space Station. The main entrance page is here, but most of the links are password protected...
Not sure why I'm not being modded up, but the link details all of the technical specs of OS and hardware, as well as the code and networking info. Shines quite a bit of light on the story... Surprised the original poster didn't look this far...
LR
...And I have a shell account, haha! Email me and I'll ask the admin about getting you one, too: spacejunk@satellite.nasa.gov
Need a Linux consultant in New Orleans?
Hmm, did I hear my name called? :-)
~~
By the way, this story is a dupe, and I don't think anyone else has called on it yet. It was posted several months ago under a topic of something like '.mars is coming.' and talked about how satellites around Mars would use TCP/IP.. just like, er, this article.
mogorific carpentry experiments
Don't be silly. You can't run Cat5 into space, it's too long of a run. This is definitely a job for single-mode fiber
So, in college, I worked with some astrophysicists at the Enrico Fermi Institute, which is where they build nuclear powered satellites, and took some classes from professors at at The Laboratory for Advanced Space Research, which is responsible for building such satellites/spacecraft as Ulysses, Pioneer 10 & 11, Cassini, StarDust, and Argos.
Anyhow, from talking with folks at the EFI and LASR, the general answer to everybody's questions is: latency and noise. Remember, this is a Cosmic Hot Interstellar Plasma Spectrometer, which means that it's going to be sent away from the earth, and eventually be millions and billions and trillions + miles away. The longer that it works, the more latency is going to build up... So, the programming needs a very non-interactive protocol. If this thing goes interstellar, it could take days and weeks for packets to travel from Earth to CHIPS and back.
Remember, it takes 4 minutes for light from our closest neighboring star to reach earth, traveling at, well, the speed of light. In all probability, this CHIPS will be using radio frequencies which are much, much slower. (I could be wrong, but I would be surprised if they had hacked some type of interstellar laser guidance system... )
Anyhow, they write scripts for this kind of mission, and generally operate with a big time lag, to the extent that it's sort of like typing with your computer monitor turned 'off'. That is, they'll figure out what they want the satellite to do for the next week or next month, type up a script, and 'submit' it to CHIPS. A couple of hours/days/weeks later, CHIPS will receive the script and start working. This kind of astrophysics programming generally involves being able to project into the future (temporally), and to know that in {x} days, the satellite will be past Mars, in {x+a} days, it'll be past the asteroid belt, in (x+a+b} days, past Jupiter, in {x+a+b+c} days, past Saturn, and so forth. It also requires good file keeping and record keeping, so that you know how many days {n} into the project you are, so you can calculate {n-x}, which gives you the time window for submiting an FTP control sequence.
Other than that, yeah... they can dir things and get thing. Depends on the exact implementation, but you have the concept.