WEBVTT

00:00:00.120 --> 00:00:05.879
between the radiation the micro gravity

00:00:03.280 --> 00:00:12.559
and the extreme constraints on Power and cooling space is just about the worst

00:00:09.160 --> 00:00:14.240
possible place to put a computer can you

00:00:12.559 --> 00:00:19.920
guys let me in that's better now if you follow space

00:00:18.039 --> 00:00:25.599
research and exploration you probably know that space is just full of

00:00:21.880 --> 00:00:28.640
computers like this one so how do they

00:00:25.599 --> 00:00:31.920
do that well the short answer is by

00:00:28.640 --> 00:00:34.120
replacing them a lot the ISS takes

00:00:31.920 --> 00:00:39.640
regular shipments of dozens of laptops at a time which get and this is great

00:00:35.920 --> 00:00:42.680
this is a direct quote absolutely

00:00:39.640 --> 00:00:45.960
destroyed but not every computer can be

00:00:42.680 --> 00:00:49.000
disposable and in 2017 the madlads at

00:00:45.960 --> 00:00:51.160
Nasa HP Enterprise and kokia who

00:00:49.000 --> 00:00:55.879
sponsored this video collaborated to create spaceborn one the first Edge

00:00:54.039 --> 00:01:00.519
Computing server that was intended to run for an extended period of time on

00:00:58.280 --> 00:01:05.400
the International Space Station of course it being their first attempt

00:01:02.600 --> 00:01:09.119
some uh let's say learning took place and it turned out that the super

00:01:07.080 --> 00:01:14.439
capacitors in the ssds were prone to radiation related failure who knew but

00:01:12.799 --> 00:01:19.920
since all it was ever meant to do was run benchmarks anyway the mission was

00:01:16.799 --> 00:01:22.560
considered a huge success and in 2021

00:01:19.920 --> 00:01:27.720
they launched spaceborn 2 whose purpose was to move Beyond proof of concept and

00:01:24.840 --> 00:01:33.240
explore practical applications for onst compute especially data analysis using

00:01:30.320 --> 00:01:38.360
AI but the story doesn't end there behind me is new space born 2 for

00:01:36.680 --> 00:01:43.000
administrative reasons it has the same name and Core specs as last time but it

00:01:40.880 --> 00:01:48.640
just took off earlier this year and it features more storage than ever over 130

00:01:46.479 --> 00:01:52.439
tabt which is an incredible feat when you consider the design challenges I

00:01:50.159 --> 00:01:58.520
mean where do they even install these things look up oh right I guess they

00:01:56.399 --> 00:02:01.840
don't really need a ladder do they no but you might

00:02:09.440 --> 00:02:16.400
to illustrate why Edge Computing is needed on the ISS let's look at a use

00:02:13.599 --> 00:02:20.680
case that's focused on astronaut safety these are the Eva gloves that the crew

00:02:18.560 --> 00:02:26.200
members wear during space walks and according to this article from 2016 they

00:02:23.120 --> 00:02:28.879
were responsible for half of all space

00:02:26.200 --> 00:02:33.400
suit injuries so to ensure their integrity between you es NASA requires

00:02:31.280 --> 00:02:37.120
the crew to take hundreds of photographs of them from every angle and then beam

00:02:35.640 --> 00:02:41.959
them back to Earth where machine learning is used to analyze them for

00:02:39.080 --> 00:02:49.080
scratches or other hazards except for one small problem that data transfer

00:02:44.680 --> 00:02:52.640
takes five days but with spaceborne oh I

00:02:49.080 --> 00:02:55.400
don't know how about 45 seconds not only

00:02:52.640 --> 00:02:59.599
is this a huge time sa but with only a handful of pictures needing to go to

00:02:57.040 --> 00:03:03.799
Earth for further analysis spaceborn can free up a significant amount of the

00:03:01.560 --> 00:03:08.760
crew's limited network bandwidth for other more interesting things with such

00:03:07.159 --> 00:03:13.519
obvious benefits then you got to be wondering why did no one ever try to put

00:03:10.879 --> 00:03:17.760
a server on the ISS before the short answer is after seeing how the crew

00:03:15.519 --> 00:03:22.280
laptops fared many people thought that they just plane wouldn't work and even

00:03:20.239 --> 00:03:27.200
if they did there were a host of other hurdles to clear like the launch okay

00:03:25.000 --> 00:03:31.879
this is really cool rocket companies like SpaceX and Northrop Grumman have

00:03:29.360 --> 00:03:35.599
Shak test machines that are programmed with profiles that will simulate the

00:03:33.799 --> 00:03:40.000
launch conditions of their respective rockets and if you've seen that viral

00:03:38.040 --> 00:03:44.120
video of the machine that disassembles hard drives by vigorously shaking them

00:03:42.439 --> 00:03:49.439
you're going to know that surviving that kind of treatment is no mean feed well

00:03:47.640 --> 00:03:54.200
these machines Managed IT both in the simulation and in the real world they

00:03:52.120 --> 00:03:59.400
actually lift it off at the end of this January every piece of equipment must

00:03:56.640 --> 00:04:03.159
also pass an acoustic chamber test and a US user friendliness evaluation to

00:04:01.280 --> 00:04:07.680
ensure the station crew can install and manage it and uh oh here's a good one

00:04:05.680 --> 00:04:13.360
apparently all equipment sent up to the ISS goes through what's called a white

00:04:10.040 --> 00:04:15.439
glove test which uh thankfully is not

00:04:13.360 --> 00:04:20.639
what it sounds like basically you put on a pair of white gloves and then you just

00:04:17.840 --> 00:04:26.400
manhandle the crap out of it if the gloves snag or tear on

00:04:23.600 --> 00:04:30.880
anything yeah that's a potential source of injury I'm going to need you to file

00:04:28.240 --> 00:04:36.000
that down which f fun fact they actually do onsite and then repeat the test I

00:04:33.919 --> 00:04:41.400
just hope they weren't filing any RAM sticks speaking of let's take a closer

00:04:38.600 --> 00:04:45.479
look at these machines machine Z cuz it's not just one in here I know it was

00:04:44.240 --> 00:04:49.080
kind of the point of this whole experiment but it still weirds me out

00:04:47.240 --> 00:04:52.520
that these are just bog standard HP Enterprise systems that you could order

00:04:50.520 --> 00:04:57.160
on their site today they don't even have lead armor or anything in this case

00:04:54.960 --> 00:05:02.600
we're looking at an edgeline 4000 which is a multi-blade system and a dl360 dual

00:05:00.199 --> 00:05:07.280
socket server we asked why these specific machines and the answer we got

00:05:04.840 --> 00:05:12.880
was shockingly relatable we sorted the HP server catalog by depth power draw

00:05:10.000 --> 00:05:18.880
and GPU support and these were the ones we were left with all right fair enough

00:05:16.000 --> 00:05:23.400
as for why two different machines well here's the thing in a perfect world

00:05:21.560 --> 00:05:28.720
multiples of the same machine would have been better but due to power constraints

00:05:26.280 --> 00:05:34.039
they chose to have one with more CPU cores for more additional scientific

00:05:30.759 --> 00:05:37.000
applications and one with pure CPU cores

00:05:34.039 --> 00:05:41.680
but with a GPU for deep learning and AI one thing they needed for both however

00:05:38.880 --> 00:05:46.039
is Ample Storage kokia generously sponsored this and brought us out here

00:05:43.520 --> 00:05:51.479
so let's take a look at the let's call them unique choices that they made for

00:05:48.280 --> 00:05:53.520
their storage configurations first up

00:05:51.479 --> 00:05:59.680
obviously gone is any trace of super capacitors so kokia can proudly say that

00:05:56.560 --> 00:06:02.319
their ssds are space ready I guess but

00:05:59.680 --> 00:06:08.360
but what's less obvious is why they chose a SAS interface Drive rather than

00:06:05.520 --> 00:06:12.680
NVMe for their high-speed bulk storage I mean you would think this is Space Age

00:06:10.680 --> 00:06:17.280
Technology they'd want the fastest thing possible but these drives were selected

00:06:15.000 --> 00:06:22.479
for their balance of performance reliability and especially power

00:06:20.319 --> 00:06:28.599
efficiency when you're looking at a shared power budget across two servers

00:06:25.919 --> 00:06:34.960
that is less than a typical gaming rig every single lot counts oh right and

00:06:32.400 --> 00:06:39.400
that's even under ideal conditions at any given time to conserve power for

00:06:37.240 --> 00:06:44.240
other priorities on the station the team can be asked to operate in half power

00:06:41.560 --> 00:06:49.880
mode or even to shut down entirely for large operations like docking so the new

00:06:46.560 --> 00:06:54.080
DL 360 server this guy right here gets

00:06:49.880 --> 00:06:55.319
four 3.72 TB pm6 Enterprise drives

00:06:54.080 --> 00:07:02.160
totaling 120 tabt of raw bulk storage for

00:06:58.960 --> 00:07:04.080
scientific data and for backups then for

00:07:02.160 --> 00:07:11.120
application drives we've got a really wild config again they went with four

00:07:06.919 --> 00:07:13.319
drives but this time it's their RM sixs

00:07:11.120 --> 00:07:18.080
again they're using SAS for lower power but this time two of the drives are

00:07:15.680 --> 00:07:21.919
operating in a data redundant mirror and the other two are basically just

00:07:20.000 --> 00:07:27.240
chilling there ready to be put into action in the event of a failure two

00:07:24.039 --> 00:07:30.199
warm spares out of a four Drive array

00:07:27.240 --> 00:07:35.039
would sound like crazy Paran oia on Earth but I assure you that in space

00:07:32.680 --> 00:07:39.800
where bit flips from random radiation are much more common it's perfectly

00:07:37.919 --> 00:07:45.199
reasonable I mean other than the overkill Drive config and the 28v power

00:07:43.360 --> 00:07:49.159
conversion that they need to run it on the ISS there's not much to say about

00:07:47.680 --> 00:07:53.400
this thing it's pretty much a bog standard server there is one cool demo

00:07:51.599 --> 00:07:58.120
that they said we could run though oh yeah they offered to let us pull one of

00:07:55.599 --> 00:08:02.479
the drives out of this dummy machine and live swap it in into this running

00:08:00.240 --> 00:08:07.240
machine to show that no data loss will occur you want to do the honors sure all

00:08:05.560 --> 00:08:13.599
they asked is that you put it in Bay 8 see you can see the drive is actually operating use Bay s that works look at

00:08:11.440 --> 00:08:18.639
that I mean that's good rm6 that would be bad if it wasn't

00:08:15.800 --> 00:08:24.080
kokia and then let's check the size beautiful 3.8 tab exactly what we want

00:08:22.039 --> 00:08:26.879
they're not the exact drives that are on the space station but we wanted a

00:08:25.360 --> 00:08:32.880
different capacity to show you that it's working and those are expensive

00:08:30.000 --> 00:08:35.959
look at that status rebuilding I mean this seems like a lot of extra steps we

00:08:34.399 --> 00:08:40.519
could have just looked up the light yeah the light yeah it's going

00:08:38.159 --> 00:08:44.839
success and whether you're looking for a SAS drive an NVMe drive high capacity or

00:08:43.159 --> 00:08:49.360
high performance we're going to have a bunch of Kyo has great Enterprise grade

00:08:47.399 --> 00:08:53.640
drives Linked In the description down below I think I'll let you take this one

00:08:51.360 --> 00:08:57.839
apart it's uh appropriately line this size okay let's take a look at the

00:08:55.600 --> 00:09:04.440
second server that's packed into each Locker the el4 ,000 is a blade chassis

00:09:01.160 --> 00:09:07.279
so the servers are basically these

00:09:04.440 --> 00:09:12.920
little slide in cards that yeah I know AR these cute or what wait pull it out

00:09:09.279 --> 00:09:15.920
and they go in on the side look at that

00:09:12.920 --> 00:09:20.279
instead of from the front wow and they

00:09:15.920 --> 00:09:24.240
managed to sneak four kokia xg6 NVMe

00:09:20.279 --> 00:09:26.240
drives into each of these blades well

00:09:24.240 --> 00:09:30.320
when I say each of these blades I should say they had the power budget for four

00:09:28.399 --> 00:09:35.680
drives but they didn't have the power budget for four blades in the flight

00:09:33.279 --> 00:09:40.200
configuration of this system they ship with just one of the four blades

00:09:38.000 --> 00:09:45.120
installed though it should be noted they do fly up a spare blade per system in

00:09:42.480 --> 00:09:49.959
the event of a failure there's just no way that that poor Locker can support

00:09:47.279 --> 00:09:55.760
both these blades and the other server running concurrently let's put you away

00:09:52.720 --> 00:09:59.000
and shift our Focus to the locker now

00:09:55.760 --> 00:09:59.880
obviously there's no real up or down on

00:09:59.000 --> 00:10:06.800
the as it whizzes around the earth at around 28,000 km an hour but to improve comfort

00:10:05.040 --> 00:10:12.640
for the astronauts they tend to mount directional items like plants in a fixed

00:10:09.959 --> 00:10:18.040
orientation which will put our lockers the drawers that hold our servers in the

00:10:15.560 --> 00:10:22.279
ceiling there are two of these lockers each containing an identical system load

00:10:20.040 --> 00:10:27.040
out for workload sharing and redundancy and these lockers present some serious

00:10:24.519 --> 00:10:32.760
design challenges starting with the fact that they use a standard that quite lit

00:10:29.279 --> 00:10:35.320
Lally doesn't exist on Earth Express

00:10:32.760 --> 00:10:39.279
rack to pack the servers in then HP Enterprise had to get kind of creative

00:10:37.600 --> 00:10:43.680
they found the shortest servers they could and then they stuffed them in

00:10:41.399 --> 00:10:48.680
sideways and they're using a combination of air Cooling and water cooling the air

00:10:46.279 --> 00:10:53.920
cooling uses a system on the ISS called AAA so at the back of these lockers

00:10:51.519 --> 00:10:59.279
there's two cold air supplies and then two hot air returns that handles about

00:10:56.600 --> 00:11:04.079
20% of the cooling for the servers op viously 20% not 100% they're going to

00:11:02.320 --> 00:11:08.480
need some more and that's where this water cooling comes in this isn't a one:

00:11:06.920 --> 00:11:15.959
one for how it would be deployed on the ISS for one thing these fittings 3D

00:11:11.320 --> 00:11:18.040
printed mockups real fitting $800 a pop

00:11:15.959 --> 00:11:23.120
if you were even allowed to buy them this tubing cheap vinyl from Home Depot

00:11:21.040 --> 00:11:26.800
real tubing must be made of stainless steel in fact any wetted surface so

00:11:25.440 --> 00:11:30.480
anything that comes in contact with water is supposed to be made out of

00:11:28.639 --> 00:11:35.839
stainless steel but we can still illustrate how it's supposed to work so

00:11:33.200 --> 00:11:39.720
on this side these go into a heat exchanger much like this one this is

00:11:37.959 --> 00:11:46.240
actually from the first generation spaceborn but functionally it's the same

00:11:42.560 --> 00:11:48.360
it pumps cold water into here chills the

00:11:46.240 --> 00:11:53.320
air inside the system and then takes the warm water out to be dissipated to the

00:11:51.040 --> 00:11:56.959
photovoltaic heat exchangers that are plumbed up with liquid ammonia coolant

00:11:55.440 --> 00:12:01.519
and mounted to the exterior of the station to sink that heat into space

00:11:59.800 --> 00:12:07.360
you need these kinds of special heat exchangers because while we think of

00:12:03.760 --> 00:12:10.720
space as cold and we see people you know

00:12:07.360 --> 00:12:13.560
oh blasted out of airlocks and they

00:12:10.720 --> 00:12:18.760
freeze over or whatever in movies the truth is that for traditional methods of

00:12:15.800 --> 00:12:25.240
heat dissipation you need air and in the near vacuum of space well it ain't

00:12:22.199 --> 00:12:27.920
there air get

00:12:25.240 --> 00:12:32.440
it cheesy jokes aside the two cooling systems together are good for removing

00:12:29.560 --> 00:12:38.800
about 400 watts of heat from each locker but that's a combined budget so if this

00:12:35.079 --> 00:12:41.560
GPU server kicks into high gear well

00:12:38.800 --> 00:12:45.680
these CPUs better just chillax for a little bit now let's talk about one of

00:12:43.519 --> 00:12:50.160
my favorite subjects networking there's four standard RJ45 ports on the front of

00:12:48.199 --> 00:12:56.519
the okay I'm going to show you on the real one ah as I was saying four ports

00:12:53.399 --> 00:12:58.839
on each Locker two of them connect both

00:12:56.519 --> 00:13:03.880
of the iss's internal gigabit networks to a separate redundant switch inside

00:13:01.440 --> 00:13:09.399
the locker and then the other two links are going to go between the two lockers

00:13:05.959 --> 00:13:11.560
at 10 GB why 10 gig well because for

00:13:09.399 --> 00:13:16.120
either backups or for multinode workloads that is a heck of a lot better

00:13:13.680 --> 00:13:21.320
than gigabit and the power budget didn't allow for anything faster cool I guess

00:13:18.519 --> 00:13:26.440
I'm starting to notice a pattern here anyway that's all pretty standard but

00:13:24.040 --> 00:13:33.680
things become less so when you look at the station side of these Networks

00:13:29.320 --> 00:13:36.040
cables this is a 37 pin military spec

00:13:33.680 --> 00:13:41.839
locking connector these are designed for power and data but in this application

00:13:38.760 --> 00:13:43.000
just eight of the pins would be used and

00:13:41.839 --> 00:13:49.079
it is $220 for just this part now on the space

00:13:46.880 --> 00:13:54.680
station NASA provides these cables for you but for testing sake really here on

00:13:51.839 --> 00:13:58.120
Earth HP Enterprise had to make their own

00:13:56.240 --> 00:14:02.079
fantastic what's really going to blow your mind though is for all of their

00:13:59.880 --> 00:14:07.079
expensive networking these machines do not have a normal internet connection

00:14:04.880 --> 00:14:14.040
just a private link back to Earth that NASA not only limits to a mighty 1

00:14:11.240 --> 00:14:20.880
megabit per second but that they also encourage folks not to make full use of

00:14:17.240 --> 00:14:24.360
also even now in 2024 it doesn't have

00:14:20.880 --> 00:14:26.360
247 connectivity pretty much every hour

00:14:24.360 --> 00:14:31.079
or two there's a period of downtime that can be anywhere as short as a few

00:14:27.759 --> 00:14:32.519
minutes or as long long as 45 minutes

00:14:31.079 --> 00:14:36.759
and that's because they have to prioritize generating enough power for

00:14:34.560 --> 00:14:41.040
the station and when the giant solar arrays Point toward the sun they can

00:14:39.000 --> 00:14:46.279
block line of sight with the satellites that provide connectivity which oh

00:14:43.639 --> 00:14:52.759
that's a fun fact even though the ISS orbits less than 500 km from the surface

00:14:49.040 --> 00:14:56.120
of the Earth our ping times to the ISS

00:14:52.759 --> 00:14:59.040
and yes we got to Ping the ISS which was

00:14:56.120 --> 00:15:04.480
pretty cool but our ping times were a atrocious reaching nearly a second as we

00:15:02.440 --> 00:15:10.560
uploaded some of the dankest memes that Earth had to offer LT store.com now we

00:15:08.800 --> 00:15:18.040
asked why that is and the answer was twofold one it's really old okay fair

00:15:15.079 --> 00:15:22.560
enough but also too the station's Internet relay is in geosynchronous

00:15:20.759 --> 00:15:28.399
orbit over 35,000 kilm from the Earth's surface and

00:15:25.720 --> 00:15:32.360
uh well what I said was H well there's your problem right there it's just

00:15:30.000 --> 00:15:36.759
really far and you might be wondering well why don't they just use

00:15:34.360 --> 00:15:41.240
starlink that's a good question um someday they might but for now they

00:15:38.880 --> 00:15:44.040
don't and HP and the team on the ISS have to work around the constraints of

00:15:42.639 --> 00:15:48.199
the current setup I mean for crying out loud it took them four years to validate

00:15:46.600 --> 00:15:51.279
that you know we can even just run a normal computer up here and actually

00:15:49.759 --> 00:15:55.360
expect this thing to be reliable they can't just switch to something and go I

00:15:53.079 --> 00:15:59.639
don't know I hope it works um oh by the way here's another fun one there's no

00:15:57.360 --> 00:16:05.199
API to determine if if their connection is up or down so instead what they do is

00:16:02.639 --> 00:16:10.240
buffer all their Communications in basic terms that means that they ping every

00:16:07.920 --> 00:16:14.600
second and if the ping succeeds they send data and then hope that the

00:16:12.560 --> 00:16:19.399
connection stayed up during that time it's a pretty good system okay not a

00:16:17.319 --> 00:16:24.000
perfect one but certainly enough for us to learn a lot from the spaceborn

00:16:21.560 --> 00:16:28.639
project even though spaceborn 2 has been in action for 3 years there's still so

00:16:26.560 --> 00:16:33.199
much to learn two of the servers have hard Hardware raid cards for their

00:16:30.240 --> 00:16:37.959
drives for example costing both mass and power consumption while two of them use

00:16:35.519 --> 00:16:42.480
software raid which obviously doesn't consume any Mass but could impact power

00:16:40.560 --> 00:16:47.680
consumption even more depending on the loads and one of them could be more or

00:16:45.480 --> 00:16:52.920
less reliable than the other we won't know until we try which is kind of a

00:16:50.600 --> 00:16:55.959
recurring theme here so if you want to learn more about the spaceborn computer

00:16:54.839 --> 00:16:59.519
project we're going to have some resources linked for you down below and

00:16:57.959 --> 00:17:04.000
we're also going to have a L to some great Enterprise storage options from

00:17:01.480 --> 00:17:08.720
our sponsor kokia we're truly grateful for this unique opportunity to get

00:17:06.280 --> 00:17:15.720
realistically as close as I ever will to the real ISS a prop in a sound stage in

00:17:13.000 --> 00:17:20.600
LA but hey thanks kogia for the opportunity and for your long-term

00:17:17.600 --> 00:17:20.600
partnership