Minutes of the DatadyneTM board meeting, May 7, 2037

Members of the board called to order by Chair Mykyta Coetzee. Chair address begins.

There is no more lucrative opportunity for our company than to create a data center with the highest security and integrity — security and integrity that can be sold to governments and corporations, individuals and academia, even the criminal underground. Everyone needs a safe. Yesteryear those safes had to hold tangible goods: diamonds, documents, bodies.... But information has been the barter of choice for decades and Earth is no longer the safest place to store data. Last month's incursions at all twenty of Google's data centers proved that all the encryption in the world cannot stop someone from simply taking the hard drives and thereafter breaking the encryption at their convenience.

Of course, what Google didn't tell the public is that the encryption didn't matter. Our sources revealed last night that that the provocateurs didn't steal anything. They set off limited EMP devices on the campuses. Devices that appear to have targeted the same data blocks at every center. Google's still trying to figure out everything that was lost. And whomever owned the data that was the target of such… thorough… espionage isn't talking. No, we need a better solution.

Since the Pinkertons we've known that security is a trade-off between convenient access and value. After all, what's the point of storing an invaluable Picasso in an impenetrable safe if the only way you can enjoy it is to schedule security on an impromptu, even random basis so you can sip a brandy and gloat over your ownership for a few minutes? Of course, information is far more transportable and its security far less reliable. Which is why I believe we should build a Terra-altiore data center somewhere only a few organizations (and even fewer people) can access.

It is proposed that an archival data center be built on Mars.

  • The center must be energy self-sufficient, which means nuclear power with either mined uranium that is locally refined or periodic shipments from Earth. The periodic shipments are preferred as they can be automated in transit with conditions that guarantee no biological life can survive the trip.

  • The center wants to have access to as much water as possible.

  • The center wants to be as geologically stable as possible.

  • The center wants to be as protected from sand storms and other weather as possible.

  • The center wants to be as protected from celestial interruption (solar storms, radiation, etc.) as possible.

  • Cost is an issue, so the center cannot be built entirely underground. Given the above requirement for geological stability, it may not be possible to build any of it underground. However, no more than 50% of the center may be considered underground for the purpose of this analysis. Indeed, while this center could easily cost in the trillions to build, respondents should not answer on the basis that cost is no object.

  • The center cannot use Mars-orbit satellites to communicate with Earth. Don't ask me why, it's simply a condition of the question. So say we all. That means land-based communication (BHAs, also known as big honking antennas) and that means the center wants the best telemetry with Earth possible.

  • The time to move data back and forth to Earth is irrelevant. The center is an archive — a place to put stuff that doesn't require immediate access.

  • Yes, the data will be well encrypted. The point of putting the site on Mars is to minimize the potential for physical disruption of the site. Partisans can be such a pain in the tuches.

  • Tech is circa 2037 (near future). Other than the ability to move materials to Mars, technology is only somewhat more advanced than today. This question is not asking about the viability or ability to build the facility. Please assume that either enough materials and equipment can be derived from Mars or can be economically shipped from Earth to make the facility happen.

Question: Your mission, should you choose to accept it, is to select a location on Mars that best suits all of the above conditions. Obviously there must be trade-offs. Therefore, suitable answers must justify their choice by explaining what trade-offs were entertained and why the answer's selection of trade-offs best support the goal of a secure and viable data center.

The best answer will maximize all the conditions for the data center.

  • $\begingroup$ if the goal is to make the data safe from the people with the resources to break into 20 of google's data centres at the same time to destroy some data what's to stop them from just dropping some rocks on your data centre from space. try using a fleet of trident submarines as data centres instead $\endgroup$
    – mgh42
    Commented Jun 17, 2019 at 4:02
  • $\begingroup$ @BilboBaggins, per my question, the time delay is irrelevant and #2 has the potential of being an answer. $\endgroup$
    – JBH
    Commented Jun 17, 2019 at 4:32
  • $\begingroup$ @mgh42, what on earth makes you think that dropping rocks on Mars is in any way as easy as inserting 20 teams on Earth? Things on Earth are easily (relatively) attacked by people on Earth. Things on Mars are not (unless you can demonstrate otherwise). $\endgroup$
    – JBH
    Commented Jun 17, 2019 at 4:34
  • 1
    $\begingroup$ It's highly unlikely that you'll be able to find viable uranium deposits on Mars. Deposition of uranium of Earth relies on an active hydrosphere, and the initial concentration of the element in the crust relies on out planet's long history of active tectonics. Neither of these are present on Mars. $\endgroup$ Commented Jun 17, 2019 at 20:38
  • 1
    $\begingroup$ No problem! In fact, a stirling generator is probably more cost efficient than a nuclear reactor for this purpose. I expect there'd be enough heat differential between the bottom of a borehole and the atmosphere to run one, and then your cost becomes the one-time delivery of a drill rig instead of regularly delivering fuel grade uranium from Earth. (Which is likely going to be deemed illegal anyway due to the risk of your delivery rocket exploding during launch and spreading radioactive material around.) $\endgroup$ Commented Jun 17, 2019 at 21:04

5 Answers 5


So, most important requirement first:

Cost is an issue, so the center cannot be built entirely underground. Given the above requirement for geological stability, it may not be possible to build any of it underground. However, no more than 50% of the center may be considered underground for the purpose of this analysis

This is of course, a load of cobblers. If cost was an issue, it would be built on the moon. Or multiple redundant orbital datacentres. Or on an abyssal plain. All of these are extremely difficult to get to, and interfacing with a secure datacentre when you've got there will clearly be beyond the abilities of all but a very small group of organisations.

The Martian option is vastly more expensive and vastly less convenient and doesn't demonstrate a usefully greater level of security than those, so clearly this is a vanity project and money is no object. Lets not pretend otherwise.

Now we've got that that out of the way, the baffling restriction on "no more than 50% of the facility may be underground" is so poorly specified that it must be the product of upper management who are both financially and technically incompetent, as even the most engineering-adverse beancounter would give a much tighter set of restrictions to minimise cost. Given that they don't know what they're talking about, this issue may be worked around to a certain extent.

The facility will be on Mars, which has a little over a third of Earth's gravity, and so little water that digging holes and burying equipment is a trivial exercise compared to shipping a datacentre there. You don't even have to do any environmental remediation; its a dead wasteland. Take a look at this PDF: Radiation Protection Strategy Development for Mars Surface Exploration (PDF). Scroll to page 12 or so, and you start seeing nice diagrams like this:

Surface plot of 365 day exposure inside ILMH – regolith shielding

2 metres of dirt is all it takes. Hell, even a metre will do it at low altitudes. In Martian gravity, that's a pretty light loading for the roof of the facility to support. You'll probably need to send out robot bulldozers to prepare the site anyway, and scraping a shallow grave for your folly is well within their capabilities. Once the facility is assembled, just bulldoze the extracted regolith back over the top in a comparatively thin layer.

I'm sure the same upper management drone will complain about how this is still technically "underground", but really all we're doing is using loose regolith as a coating material... it is no more "underground" than a stone house with a slate roof is somehow "underground" because of what it is made of. I'm sure you'll be able to give them the brush-off, especially when you show them the figures for the reduced cost of shipping shielding from Earth.

The center cannot use Mars-orbit satellites to communicate with Earth.

I have actually sprained my eyes they're rolling so hard. You need to find the pointy-haired imbecile who is clearly trying to sabotage this project and fire them (and at the very least, replace them with someone who knows how to properly derail it rather than making these half-hearted bizarre requirements that can be worked around with a little bit of thought).

You'll be happy to know that my backup plans involve communication mechanisms that can be mounted to Phobos and/or Deimos, and assuming that will be rejected too (because you need to let the PHB reject something to make them feel like they're actually contributing) there's also a proposal with a more conventional constellation of communications satellites than can be placed in halo orbits around the Mars-Sun L1 and L2 Lagrange points to provide continuous coverage of the Martian surface (additional details may be found here: Sun-Mars libration points and Mars mission simulations (PDF))

SUm-Mars L1 and L2 halo orbit constellation

Nothing need orbit Mars itself. Everything will be great, budget approval will be straightfoward. Working with other national space agencies to co-develop and co-fund this part of the project is clearly possible as it has actual use unlike the main subject of this project.

The budget and design time saved by replacing the Big Honking Antennae with Sensibly Sized Antennae can help fund a more sophicated set of construction robots to facilitate the improved shielding proposal above, and provide a means to transport ice to the facility from elsewhere on the Martian surface. Having communication capability across the entire Martian surface allows the facility to be sited anywhere (not merely places with reasonably frequent lines of sight to Earth), and allows easier teleoperation of construction and support robots.

The improvement in comms link availability (24 and a bit hours a day, 687 days a year, no daily downtime) will increase facility reliability and usefulness and will help drive costs down and customer adoption rates up. The risk of downtime due to satellite failure is low (both satellites would have to go down, vs. the risk of the BHA array failing) and the logistical requirements of running the facility are sufficiently high that bringing backup satellites in shielded coccoons for deployment after a serious CME event would seem to be within the realms of practicality.

The center wants to have access to as much water as possible.

Easy. Don't build it on Mars.

Given that you've already taken the hardest, silliest approach (I mean, if you're being this mad, at least take it to an ice moon or something) then perhaps you'd like to specify what the water will be needed for? You're certainly not going to be doing anything utterly insane like open cycle cooling of your nuclear reactor with steam towers, like you're still on earth, are you? No. You are not. Similarly, the coolant loops within your datacentre are closed systems. You won't be venting steam or leaking water. If you are leaking, then you're not fit to run a datacentre in a city on Earth, let alone anywhere harder to send maintenance folk. Lets be honest, someone accidentally left in a proposal for a swimming pool, didn't they? Just quietly scratch that bit out, and your outrageous water requirements will be rendered manageable.

Now you can site the damn thing almost anywhere, and have robotic rovers go bring you some ice to fill your coolant loops. Even the driest parts of the equator have potential water access available; you don't need to park right on an icecap (and indeed probably wouldn't want to as the weather tends to be poorer there).

The possibility of good building sites coinciding with good sources of water might lead you to consider siting the facility in the northern dune sea... that's fine and the low altitude means a reduced shielding requirement, but the facility will be rendered more vulnerable to dust storms compared to high-altitude sites.

The center wants to be as geologically stable as possible.

Well, Mars would seem to be a good choice for that; Marsquakes are rare, and weak. So long as the facility isn't built on the fringes of the icecaps, or in small craters, it will likely be just fine.

The center wants to be as protected from sand storms and other weather as possible.

The Shallow Grave proposal provides excellent protection to the facility from what little threat the martian elements provide. The removal of the BHA from the plan makes protecting surface equipment fairly trivial, too.

A small ground of sweeper robots may be required in the aftermath of a dust storm, and possibly during one. Although the facility does not require solar power, backup systems may still use it and solar panels will need to be cleaned. More importantly, any heat radiators for the facility's power plant(s) and computers will need to remain fully operational. If the dust forms a thick enough layer to insulate the radiators, they will need to be cleaned off.

The possibility exists of siting the facility on Tharsis; it is known that the highest mountains are above the level of global dust storms. The shielding requirements are higher, but 2m of regolith will be sufficient.

Olympus Mons standing above a global dust storm

Olympus Mons photographed by Mariner 9, standing above a global dust storm.

The center wants to be as protected from celestial interruption (solar storms, radiation, etc.) as possible.

The shielding proposal above was designed to protect humans (known to be quite sensitive to radiation) from serious and sustained radiation from the sun and from extrasolar space. It will be quite sufficient for this project's needs.

Project management appear to be entirely disinterested in the potentially more serious issue of how to protect the facility from meteorite strike. This underlines my suspicion that this is a vanity project and isn't intended to be practical or sustainable for the long term. A deeper underground data storage facility may be required, but happily this need not be of excessive size and so would fit within the spirit and the wording of the "no more than 50% underground" requirement. Reconstructing the upper portion of the facility (including the communication arrays and cooling systems) after a meteorite strike is apparently outside of the scope of this brief, and will be left to some other poor schmuck. Please note that destruction of the reactor cooling apparatus might not be fatal to the reactor if there was emergency cooling to let the system shutdown cleanly, but a backup plan may need to include shipping out a new reactor if the old one is seriously damaged. I don't like the look of those backup cost figures, let me tell you.

The center must be energy self-sufficient, which means nuclear power

I'm... actually astonished. Presumably the previous PHB was on a coffee break or something and was unable to demand wind or solar power. Please expedite the work on this part of the project before they notice and cancel it; sink sufficient money into it and its inertia will hopefully carry us through this risky period.

A particularly large datacentre will be using less than a gigawatt of power. A nuclear reactor with an output of about 1GWe (electrical, implying perhaps 3GW thermal output) will require about 25 tonnes of enriched uranium a year. The fuel assemblies for these will of course increase the weight somewhat... using the Finnish Olkiluoto reactors as an example, the assemblies add perhaps an additional 50% overhead.

The total freight budget for the project is outside of the current set of requirements, but 50 tonnes a year sounds like it might be a reasonable ballpark figure (to include new fuel assemblies, consumables, lubricants, spare parts that cannot be fabricated on site, secure data shipment and so on). The Earth-Mars synodic period is a little over 2 years, so each actual shipment will need to take over 100 tonnes unless high-energy faster transit options are available. This will add a non-trivial continuous cost.

Smaller reactors will of course require less fuel, but will in turn limit the size of the datacentre. The scale of the operation was unspecified, so the tradeoffs here will have to be left to upper management, god help us.

The practicalities of establishing a fuel reprocessing facility on Mars, or launching highly enriched fuel rods from earth, or mining Martian uranium (we know there is some but disposition of decent ore bodies remains unknown) to produce fuel assemblies in-situ have been considered, and were rejected as being too ridiculous, even for this project.


If you don't want to go underground, the second best bet would be on the 80° North latitude line. Here's why:

Here is a map of cosmic radiation:enter image description here

Notice the 2 places where there is dark blue. One is the Hellas Basin and the other is around 70-80° North. Why not build in Hellas Basin? It is the origin of the global Martian dust storms.
Also, if you watch the video on the link, it shows that the storms generally don't reach above about 60° North.

Concerning access to water, here is a quote from Wikipedia about the Vastitas Borealis region: "In 2005 the European Space Agency's Mars Express spacecraft imaged a substantial quantity of water ice in a crater in the Vastitas Borealis region. The environmental conditions at the locality of this feature are suitable for water ice to remain stable."

The one drawback is you will have to build your BHAs below 64.81° North, or during the Martian winter, the rotation angle of Mars will prevent them from transmitting to Earth. However, above 64.81°, you will have 24/7 reception during the summer.

The place that generally avoids the storms and has the second highest solar radiation protection is about 80° North.

  • $\begingroup$ Are you certain about the 24/7 visibility from earth? Sounds slightly surprising. $\endgroup$ Commented Jun 17, 2019 at 14:35
  • 1
    $\begingroup$ Yes, unless Mars is eclipsed by some other astronomical body. 64.81° North is the Martian Arctic circle, and since Earth's orbit is inside Mars', The antennas will always be visible in the Martian summer. $\endgroup$ Commented Jun 18, 2019 at 4:08
  • $\begingroup$ Interesting, I hadn't considered that. $\endgroup$ Commented Jun 18, 2019 at 6:10
  • $\begingroup$ @BilboBaggins Earth is going to be on the other side of the Sun for some of that time though... $\endgroup$ Commented Jun 18, 2019 at 15:48
  • $\begingroup$ Yes, that's why I said "eclipsed by some other astronomical body." $\endgroup$ Commented Jun 18, 2019 at 18:29

Still underground

You need to avoid solar radiation so it needs to be completely underground. The idea place is in the Martian lava tubes

They already exist thus you don't need to excavate, just build inside them. They are the proposed place for the first Martian base. In theory you could build an airlock and install atmosphere. They've been around for millions of year so are clearly stable.

In fact it would be cheaper to build completely in the tubes than above ground as you don't need to shield them from radiation.

  • 1
    $\begingroup$ Except that I said you can't put it all underground. There are trade-offs. What location maximizes viability (even if viability isn't perfect)? $\endgroup$
    – JBH
    Commented Jun 17, 2019 at 4:35
  • 3
    $\begingroup$ You stated that cost was the reason not to be underground as was geological stability. Lava tubes bypass both requirements. You don't need to excavate and it's cheaper to build in the tubes. Whats the trade off? $\endgroup$
    – Thorne
    Commented Jun 17, 2019 at 5:08

Ladies and gentlemen of the board: There is no data on earth that could possibly justify this cost.

If you want a datacenter to be well protected from attack then put it in a pressure-compensated subsea enclosure. Put it on an oil platform inspection vessel also configured to lay subsea fiber optic and power cables to an undisclosed location in the deep ocean. Like, 3,000 meters deep - utterly inaccessible even to modern Submarines as it is well beneath their crush depth. Lower it with a subsea crane onto the ocean floor and use ROVs to connect the cables. Voila, you have a secure data center. If you suspect inside sabotage (as seems likely if all 22 other centers were hit) then lay out 50 of them but only make 2 or 3 of them real and cluster them so even if you destroy one the other 2 stay up and the data is intact.

Hell... for that kind of cost, make all 50 of them real and you still aren't even at the cost of putting 1 data center on the moon. Make 500 of them and you're still cheaper than putting this on Mars.

If you want in on Mars, without Satellites then you also have to build a Martian equivalent of NASA's DSN: 3 networked big honking antennas each covering a 120 degree arc of the planet to communicate with Earth's DSN. The bandwidth will be worse than the slowest dialup you have ever used or heard of. The broadcasts from Mars to Earth can be received by any listening station and decrypted given enough time or the actual cipher. Mars is a bad idea unless there's enough people and infrastructure on Mars that they need one for themselves.

Microsoft has already tested something like this, but in shallower water. More things can go wrong at greater depths but if it's well engineered enough it will work.

  • 1
    $\begingroup$ Regarding "worse than the slowest dialup you've ever used"... the Mars Reconnaisance Orbiter gets 0.5 - 4 megabits per second with modern day tech without too much emphasis on massive continuous bandwidth. If that's your idea of "the worst dialup" then you may consider yourself fortunate. $\endgroup$ Commented Jun 18, 2019 at 6:36
  • 1
    $\begingroup$ (but +1 for everything else) $\endgroup$ Commented Jun 18, 2019 at 6:37
  • $\begingroup$ Okay but Mars Orbiter is communicating with 1 host. How slow would that become with 100 hosts? With 1,000? $\endgroup$ Commented Jun 18, 2019 at 8:17
  • 1
    $\begingroup$ Bandwidth use doesn't necessarily scale linearly with the number of users, but that's almost beside the point because the latency is so long. You queue up your requests and wait your turn, and eventually your pointlessly expensive stuff comes back. In that respect at least it is a little more like a postal operation; slow, but not super low bandwidth by any means. $\endgroup$ Commented Jun 18, 2019 at 8:21

Distributed blimp network.

The usual way around these sorts of problems is to stop putting your eggs in one basket and to put a few eggs in lots of baskets, then make them constantly synchronize with each other.

You'll put lots of solar panels on the balloons, and maybe some thermal radioactives.

If you still think you need water for some reason, it could be extracted (slowly) from the atmosphere. Snowfall might help a bit.

Conveniently, this sort of scheme can be arranged to avoid an enormous up-front deployment cost and spread iterative deployments into ongoing operational costs.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .