[{"author": "\u00c9ric Vyncke", "text": "

is audio ok for remote

", "time": "2026-03-17T08:01:47.000Z"}, {"author": "\u00c9ric Vyncke", "text": "

?

", "time": "2026-03-17T08:01:53.000Z"}, {"author": "Nishanth Sastry", "text": "

yes, works, thanks

", "time": "2026-03-17T08:01:59.000Z"}, {"author": "Scott Burleigh", "text": "

A little faint, but we can hear.

", "time": "2026-03-17T08:02:13.000Z"}, {"author": "Colin Perkins", "text": "

Joerg should speak closer to the mic

", "time": "2026-03-17T08:02:46.000Z"}, {"author": "Andrew Campling", "text": "

It helps if the Chairs eat the mic when they speak

", "time": "2026-03-17T08:03:02.000Z"}, {"author": "Colin Perkins", "text": "

Thank you!

", "time": "2026-03-17T08:03:12.000Z"}, {"author": "Scott Burleigh", "text": "

Thanks, better.

", "time": "2026-03-17T08:03:14.000Z"}, {"author": "Juan Fraire", "text": "

There it goes

", "time": "2026-03-17T08:03:14.000Z"}, {"author": "Jorge Amodio", "text": "

better audio now

", "time": "2026-03-17T08:04:49.000Z"}, {"author": "Dan York", "text": "

Just to be clear, Blue Origin is a separate company from Amazon. Owned by the same person, Jeff Bezos ... but a separate company.

", "time": "2026-03-17T08:09:17.000Z"}, {"author": "Dan York", "text": "

I don't know that it has anything to do with the IETF/IRTF, but if someone could do some research on the actual viability on data centers in space, that could be super useful. I've seen different commentary about how it won't work as well as it is currently being hyped.

", "time": "2026-03-17T08:14:34.000Z"}, {"author": "Joerg Ott", "text": "

We may be able to get a talk into this direction for Vienna, let's see

", "time": "2026-03-17T08:15:11.000Z"}, {"author": "Tony Li", "text": "

If this was IETF, I would ask to see running code, i.e., if it's not deployed, then it's marketing.

", "time": "2026-03-17T08:16:26.000Z"}, {"author": "Brian Trammell", "text": "

it'd be neat to hear from actual rocket scientists in that conversation.

", "time": "2026-03-17T08:16:34.000Z"}, {"author": "Nishanth Sastry", "text": "

Economist last week: \"Space Data Centers: Less crazy than you think\": https://www.economist.com/science-and-technology/2026/03/02/data-centres-in-space-less-crazy-than-you-think

", "time": "2026-03-17T08:17:06.000Z"}, {"author": "Brian Trammell", "text": "

(my understanding is the economics only make sense if you assume infiintely reliable (or disposable) hardware and the fuel is free, but I'll read the Economist article, thanks! i suspect it's somewhere in my print backlog)

", "time": "2026-03-17T08:18:38.000Z"}, {"author": "Nishanth Sastry", "text": "

(that article is good, but still has several unknowns not fully considered)

", "time": "2026-03-17T08:19:42.000Z"}, {"author": "Brian Trammell", "text": "

none of this detracts from how interesting this is at worst as a thought experiment though. this is in part what the irtf is for
\n:)

", "time": "2026-03-17T08:19:43.000Z"}, {"author": "Nishanth Sastry", "text": "

indeed :)

", "time": "2026-03-17T08:19:53.000Z"}, {"author": "Brian Trammell", "text": "

yeah it's a sci-tech article in the economist, that's house style

", "time": "2026-03-17T08:20:03.000Z"}, {"author": "Brian Trammell", "text": "

clap clap clap

", "time": "2026-03-17T08:21:07.000Z"}, {"author": "Andrew Campling", "text": "

:clap:

", "time": "2026-03-17T08:21:33.000Z"}, {"author": "Michael Richardson", "text": "

Brian Trammell said:

\n
\n

none of this detracts from how interesting this is at worst as a thought experiment though. this is in part what the irtf is for
\n:)

\n
\n

I for one, am looking forward to defending humanity against digital space aliens! (cf: \"Rapture of the Nerds\")

", "time": "2026-03-17T08:22:46.000Z"}, {"author": "Rodney Van Meter", "text": "

I'm not normally the curmudgeon type, but can someone explain to me in simple terms the motivation for data centers in space?
\nThey're expensive to launch, limited in volume, subject to serious radiation, hard to cool, limited in power, impossible to maintain, overhead only briefly if they're in LEO, high latency if they're in GEO...

\n

So what's the win? Why do this?

", "time": "2026-03-17T08:23:24.000Z"}, {"author": "Joerg Ott", "text": "

One interesting question is also keeping GPUs up to date

", "time": "2026-03-17T08:23:57.000Z"}, {"author": "Rodney Van Meter", "text": "

(Sat-based networking is a different matter. It has obvious huge wins, though as an astrophotographer I also see (quite literally) some drawbacks.)

", "time": "2026-03-17T08:24:36.000Z"}, {"author": "Nishanth Sastry", "text": "

We organised a Dagstuhl Seminar on connected space early Feb (https://www.dagstuhl.de/seminars/seminar-calendar/seminar-details/26062). Pretty much came to the same conclusion - DC in space useful for the space data, space computing category the previous speaker talked about. Other cases have too many maintanence issues and questions about how to do several basic things.

", "time": "2026-03-17T08:26:08.000Z"}, {"author": "Rodney Van Meter", "text": "

If you're talking about interplanetary probes, being able to send supercomputer-class computational capability is huge.

", "time": "2026-03-17T08:27:05.000Z"}, {"author": "Rodney Van Meter", "text": "

And some of our students are working with JAXA folks on translunar and interplanetary networks, which is awesome.

", "time": "2026-03-17T08:27:40.000Z"}, {"author": "Juan Fraire", "text": "

The space data center topic went hot after Elon's FCC filing for 1M space elements. The core arguments were sunlight availability (permanent, and more efficient above the atmosphere) and unlimited scalability (no need for surface occupation with solar arrays). The drawbacks are endless. Therefore, it is a topic that need some light from the research community.

", "time": "2026-03-17T08:28:20.000Z"}, {"author": "Nishanth Sastry", "text": "

(cue hundreds of papers and a new cottage industry in academic publishing :))

", "time": "2026-03-17T08:28:46.000Z"}, {"author": "Rodney Van Meter", "text": "

One word: Wintermute

", "time": "2026-03-17T08:29:16.000Z"}, {"author": "Michael Richardson", "text": "

any router moving at orbital speeds is a \"high-speed\" router :-)

", "time": "2026-03-17T08:31:02.000Z"}, {"author": "Rodney Van Meter", "text": "

:-)

", "time": "2026-03-17T08:31:20.000Z"}, {"author": "Nishanth Sastry", "text": "

:) but - earth is also moving at 30 metres/sec ;)

", "time": "2026-03-17T08:31:49.000Z"}, {"author": "Nishanth Sastry", "text": "

so all our routers in grubby basements too are high speed

", "time": "2026-03-17T08:32:05.000Z"}, {"author": "Rodney Van Meter", "text": "

Cramming as much computation as you can into a sensing satellite or a comm satellite makes sense. But the idea of putting DCs in space simply to get them off the ground, I don't see any compelling reason that makes dealing with those challenges anywhere close to a worthwhile effort

", "time": "2026-03-17T08:32:20.000Z"}, {"author": "Dan York", "text": "

Found one of the articles I'd seen critiquing data centers in space: https://taranis.ie/datacenters-in-space-are-a-terrible-horrible-no-good-idea/

", "time": "2026-03-17T08:34:23.000Z"}, {"author": "Brian Trammell", "text": "

there are a few things that drive it IMO, the big one being the fact there there is little penalty in Valley culture for inspiring people to do infeasible things, you might invent something profitable along the way.

", "time": "2026-03-17T08:35:02.000Z"}, {"author": "Andrew Campling", "text": "

A new IETF modus operandi: \"Running code and orbiting hardware\"

", "time": "2026-03-17T08:35:16.000Z"}, {"author": "Nishanth Sastry", "text": "

:)

", "time": "2026-03-17T08:35:27.000Z"}, {"author": "Jorge Amodio", "text": "

it is running code on orbitting hardware

", "time": "2026-03-17T08:35:53.000Z"}, {"author": "Dan York", "text": "

@Rodney Van Meter - those hyping orbital data centers seem to focus on \"infinite power\" from the sun (at least while they are not behind the Earth)... plus I wonder if there is the interest of not having to deal with all those pesky terrestrial regulations...

", "time": "2026-03-17T08:37:14.000Z"}, {"author": "Brian Trammell", "text": "

^ what dan said

", "time": "2026-03-17T08:37:30.000Z"}, {"author": "Scott Burleigh", "text": "

Rough consensus unnecessary!

", "time": "2026-03-17T08:37:34.000Z"}, {"author": "Jorge Amodio", "text": "

^

", "time": "2026-03-17T08:38:21.000Z"}, {"author": "Dan York", "text": "

Plus... you get to file with the FCC with very large numbers like 1,000,000 and freak everyone out! (and anger all the astronomers)

", "time": "2026-03-17T08:38:37.000Z"}, {"author": "Brian Trammell", "text": "

does anyone have a pointer to the work on the thermodynamics of orbital DCs? \"space is cold\" as a cooling strategy is based on terrestrial intiutions about temperature that i don't think hold...

", "time": "2026-03-17T08:38:46.000Z"}, {"author": "Jorge Amodio", "text": "

you need to do some work about cooling

", "time": "2026-03-17T08:39:19.000Z"}, {"author": "Dan York", "text": "

@brian - the article I linked to above pointed out this heat dissipation issue was the big killer of the whole idea

", "time": "2026-03-17T08:39:33.000Z"}, {"author": "Jorge Amodio", "text": "

just being in space does not do it

", "time": "2026-03-17T08:39:36.000Z"}, {"author": "Joerg Ott", "text": "

one recent blog post mentioned 1M satellites would need cooling twice the size of Manhattan, which comes down to 120sqm per sat at current tech (according to this post)

", "time": "2026-03-17T08:39:55.000Z"}, {"author": "Rick Taylor", "text": "

Not a lot of convection in a vacuum

", "time": "2026-03-17T08:39:59.000Z"}, {"author": "Rodney Van Meter", "text": "

bingo

", "time": "2026-03-17T08:40:04.000Z"}, {"author": "Rick Taylor", "text": "

There called \"radiators\" for a reason

", "time": "2026-03-17T08:40:20.000Z"}, {"author": "Joerg Ott", "text": "

What you can do is radiate (and avoid being heated by the sun at the same time)

", "time": "2026-03-17T08:40:21.000Z"}, {"author": "Brian Trammell", "text": "

right, something something infrared because the heat capacity of a vacuum is... well, it's the only way to win Oxygen Not Included

", "time": "2026-03-17T08:40:22.000Z"}, {"author": "Dan York", "text": "

Yep! You only have radiation to get rid of heat.

", "time": "2026-03-17T08:40:37.000Z"}, {"author": "Brian Trammell", "text": "

joerg: so you get a big flat satellite with solar panels on one side and radiators on the other, shifting sunlight into the IR?

", "time": "2026-03-17T08:42:17.000Z"}, {"author": "Joerg Ott", "text": "

@Brian: I don't propose this, I am wondering

", "time": "2026-03-17T08:42:46.000Z"}, {"author": "Rodney Van Meter", "text": "

Total solar irradiance is about 1.3kW/m^2. At 30% efficiency for solar cells, that's about 4m^2 per kW delivered. If you want a 1MW DC (modest size), you need 4,000 m^2 of solar cells. Big, but not as bad as I thought before doing the math.

", "time": "2026-03-17T08:42:53.000Z"}, {"author": "Brian Trammell", "text": "

put enough hydrazine on 'em for maneuvering and you can also use them as a geoengineering platform

", "time": "2026-03-17T08:42:59.000Z"}, {"author": "Tony Li", "text": "

Another aspect that I've not seen addressed is the AIDC network. How do you replicate that in space? ISLs don't cut it.

", "time": "2026-03-17T08:43:00.000Z"}, {"author": "Jorge Amodio", "text": "

Plus depending on location you have also to deal with radiation and nasty high speed cosmic particles

", "time": "2026-03-17T08:43:34.000Z"}, {"author": "Tony Li", "text": "

LEO avoids many of those issues.

", "time": "2026-03-17T08:43:57.000Z"}, {"author": "Jorge Amodio", "text": "

many but not all

", "time": "2026-03-17T08:44:19.000Z"}, {"author": "Jorge Amodio", "text": "

plus the probability of getting hit by space debires is much higher at LEO

", "time": "2026-03-17T08:44:45.000Z"}, {"author": "Brian Trammell", "text": "

\"planetlab\" is an amazing pun, 10/10

", "time": "2026-03-17T08:44:45.000Z"}, {"author": "Dan York", "text": "

But LEO has that other issue that sats typically only last 5-7 years before they run out of propellent to keep them up there.

", "time": "2026-03-17T08:45:28.000Z"}, {"author": "Dan York", "text": "

Plus Kessler Syndrome, etc., etc....

", "time": "2026-03-17T08:45:42.000Z"}, {"author": "Jorge Amodio", "text": "
                                                          and crowded orbits\n
", "time": "2026-03-17T08:46:08.000Z"}, {"author": "Brian Trammell", "text": "

5-7 years is less than the half life of an inadequately cooled rack full of GPUs though so no problem

", "time": "2026-03-17T08:46:11.000Z"}, {"author": "Warren Kumari", "text": "

I might have missed it -- if I want to contribure a node, what all is required?

", "time": "2026-03-17T08:46:12.000Z"}, {"author": "Rodney Van Meter", "text": "

Hear the onboard AI scream as it reenters the atmosphere

", "time": "2026-03-17T08:46:15.000Z"}, {"author": "Warren Kumari", "text": "

Is this just \"a VM behind Starlink?\", or a physical machne, or a rack of gear, or... ?!

", "time": "2026-03-17T08:46:36.000Z"}, {"author": "Dan York", "text": "

Nisanth - This is great work to collect all these tools!

", "time": "2026-03-17T08:46:41.000Z"}, {"author": "Jorge Amodio", "text": "

plus crowded orbits

", "time": "2026-03-17T08:46:44.000Z"}, {"author": "Brian Trammell", "text": "

very cool stuff

", "time": "2026-03-17T08:47:38.000Z"}, {"author": "Dan York", "text": "

Awesome!

", "time": "2026-03-17T08:47:45.000Z"}, {"author": "Brian Trammell", "text": "

first time i ever wished i was on satellite internet tbh

", "time": "2026-03-17T08:48:04.000Z"}, {"author": "Warren Kumari", "text": "

Cool. I have a location (which I only visit once every 3-4 months, so it is largely unsed and so less congestion) and even an unused NUC there..

", "time": "2026-03-17T08:48:38.000Z"}, {"author": "Warren Kumari", "text": "

@Brian: Move to Bluemont, VA... we have no broadband there, and so Starlink is your only option :-P

", "time": "2026-03-17T08:49:21.000Z"}, {"author": "Nishanth Sastry", "text": "

fantastic - I'll get in touch with you Warren.

", "time": "2026-03-17T08:49:36.000Z"}, {"author": "Warren Kumari", "text": "

Ta!

", "time": "2026-03-17T08:49:43.000Z"}, {"author": "Warren Kumari", "text": "

Actually, Geoff Huston used to use my network / NUC there for his Starlink latency measureents...

", "time": "2026-03-17T08:50:18.000Z"}, {"author": "Nishanth Sastry", "text": "

OK! I have seen many of his fantastic talks and graphs - would be great to have that node on LEOScope :)

", "time": "2026-03-17T08:50:58.000Z"}, {"author": "Warren Kumari", "text": "

@Brian: 'tis yer own fault for living in a first world county... \u00af\\_(\u30c4)_/\u00af

", "time": "2026-03-17T08:51:26.000Z"}, {"author": "Jorge Amodio", "text": "

some constellations have multiple orbital planes at different altitudes

", "time": "2026-03-17T08:53:20.000Z"}, {"author": "Brian Trammell", "text": "

:D i mean, I could stick a terminal on the roof for tertiary backup internet, but that seems excessive

", "time": "2026-03-17T08:53:28.000Z"}, {"author": "Joerg Ott", "text": "

@Warren: we might want to chat as well

", "time": "2026-03-17T08:53:32.000Z"}, {"author": "Warren Kumari", "text": "

Sure.

", "time": "2026-03-17T08:55:34.000Z"}, {"author": "Michael Richardson", "text": "

YAML is <-> JSON/CBOR/CDDL. Maybe describe this format in CDDL.

", "time": "2026-03-17T08:56:31.000Z"}, {"author": "Tony Li", "text": "

This ain't research, this is engineering.

", "time": "2026-03-17T08:59:19.000Z"}, {"author": "Dan York", "text": "

This is useful work!

", "time": "2026-03-17T08:59:53.000Z"}, {"author": "Tony Li", "text": "

I just said that.

", "time": "2026-03-17T09:00:10.000Z"}, {"author": "Michael Richardson", "text": "

Tony Li said:

\n
\n

This ain't research, this is engineering.

\n
\n

True. The description format is engineering... which the researchers need to drive their routing protocol simulations :-)

", "time": "2026-03-17T09:01:29.000Z"}, {"author": "Joerg Ott", "text": "

It facilitates research (reproducibility, etc.)

", "time": "2026-03-17T09:01:30.000Z"}, {"author": "Dan York", "text": "

It reminds me of some work happening to define a description format for terrestrial fiber - OFDS - https://ofds.info/en/about/

", "time": "2026-03-17T09:04:02.000Z"}, {"author": "Tony Li", "text": "

No doubt, but it would seem like it's sufficiently solid that it should be in a WG and headed towards an RFC soon.

", "time": "2026-03-17T09:04:25.000Z"}, {"author": "Juan Fraire", "text": "

Thanks for the feedback, Tony and Dan, and Michael, fully taken!

", "time": "2026-03-17T09:06:20.000Z"}, {"author": "Dan York", "text": "

Which working group would take on something like that?

", "time": "2026-03-17T09:06:26.000Z"}, {"author": "Dan York", "text": "

@Juan - it's very cool work!

", "time": "2026-03-17T09:06:47.000Z"}, {"author": "Pierre Francois", "text": "

@juan Extend it to support to non-earthly constellations? :)

", "time": "2026-03-17T09:07:13.000Z"}, {"author": "Tony Li", "text": "

I'm tempted to say NETMOD, but it's not my area.

", "time": "2026-03-17T09:07:28.000Z"}, {"author": "Dan York", "text": "

DISPATCH! :slight_smile:

", "time": "2026-03-17T09:07:56.000Z"}, {"author": "Rodney Van Meter", "text": "

See, this one is \"hey, let's process where the data is collected and reduce comm needs\" -- that's a win for GPUs in Spaaaaace.

", "time": "2026-03-17T09:08:15.000Z"}, {"author": "Dan York", "text": "

Yes! And to me this makes total sense! As he noted in his slides, the space-to-ground connection is the biggest bottleneck.

", "time": "2026-03-17T09:09:08.000Z"}, {"author": "Joerg Ott", "text": "

I don

", "time": "2026-03-17T09:09:42.000Z"}, {"author": "Tony Li", "text": "

Someone doesn't understand the word 'vacuum'

", "time": "2026-03-17T09:09:59.000Z"}, {"author": "Rodney Van Meter", "text": "

Exact ratio will doubtless vary, but I think every sensing sat and interplanetary probe we've ever sent has been comm-limited

", "time": "2026-03-17T09:10:03.000Z"}, {"author": "Joerg Ott", "text": "

Voyager certainly is :-)

", "time": "2026-03-17T09:10:29.000Z"}, {"author": "Dan York", "text": "

At the Dagstuhl seminar Nisanth mentioned, someone from... (NASA? JPL? ESA?) ... said that they are NOT able to get down more than a fraction of the images they get from their moon or Mars rovers.

", "time": "2026-03-17T09:10:39.000Z"}, {"author": "Rodney Van Meter", "text": "

Ed Stone was my PH1a prof. Fall 1982.

", "time": "2026-03-17T09:10:51.000Z"}, {"author": "Gorry Fairhurst", "text": "

Someone mentioned an IETF home for some of this?: I note that knowing we can write a spec, isn't the same as knowing people wish to implement and then use a common spec.

", "time": "2026-03-17T09:11:03.000Z"}, {"author": "Juan Fraire", "text": "

@Pierre: It would not be super crazy to also specify the central element of the orbits. The link description across planets would be overkill though...

", "time": "2026-03-17T09:11:10.000Z"}, {"author": "Pierre Francois", "text": "

Yeah, I was just thinking constellation around other bodies

", "time": "2026-03-17T09:11:40.000Z"}, {"author": "Gorry Fairhurst", "text": "

@Dan - e.g., that's true for space imaging sensors

", "time": "2026-03-17T09:12:13.000Z"}, {"author": "Dan York", "text": "

Yes!

", "time": "2026-03-17T09:12:21.000Z"}, {"author": "Nishanth Sastry", "text": "

@Dan - at the Dagstuhl, that was Dan Crichton from JPL - the farther you go the weaker the link, so local compute in space suddenly becomes viable again

", "time": "2026-03-17T09:14:01.000Z"}, {"author": "Dan York", "text": "

(I find myself wishing MeetEcho had a thumbs-up or other signal that you could do on people's messages.)

", "time": "2026-03-17T09:16:05.000Z"}, {"author": "Nishanth Sastry", "text": "

:)

", "time": "2026-03-17T09:16:21.000Z"}, {"author": "Joerg Ott", "text": "

(that would be nice indeed -- feature request?)

", "time": "2026-03-17T09:16:33.000Z"}, {"author": "Andrew Campling", "text": "

:+1:

", "time": "2026-03-17T09:16:46.000Z"}, {"author": "Dan York", "text": "

I feel like some of the LEO constellation operators think all these calculations are way too complex... it's way easier for them just to launch a zillion sats up into LEO.

", "time": "2026-03-17T09:23:02.000Z"}, {"author": "Joerg Ott", "text": "

Bigger is better?

", "time": "2026-03-17T09:23:27.000Z"}, {"author": "Nishanth Sastry", "text": "

In the end, economics is king even for the rich Elons of the world. Maybe they were just not clever enough to think of this..

", "time": "2026-03-17T09:24:10.000Z"}, {"author": "Joerg Ott", "text": "

Redundancy and safe bets?

", "time": "2026-03-17T09:24:24.000Z"}, {"author": "Dan York", "text": "

Yes, if you flood LEO with sats, you've got lots of redundancy. Plus... you get more capacity per cell.

", "time": "2026-03-17T09:27:05.000Z"}, {"author": "Dan York", "text": "

Or... as SpaceX is looking at... you launch bigger sats (v3) that have more propellant and can last longer in orbit (or do more maneuvering).

", "time": "2026-03-17T09:31:04.000Z"}, {"author": "Dan York", "text": "

Thank you J\u00f6rg !

", "time": "2026-03-17T09:32:11.000Z"}, {"author": "Gorry Fairhurst", "text": "

Thanks all - see you in Vienna

", "time": "2026-03-17T09:32:25.000Z"}]