How do I make a server of a virtual world indestructible?

Question

Imagine if we would hit a point in time where we would accept that living in a virtual world is better for being happy and would make us near immortal.

And then we would build something like the matrix and upload our minds into a virtual world.

I don't imagine it like in the movie “Matrix” but more like our brains would be uploaded like a program.

And even if our body dies we would still live as digital form in some computer/program, if we had some large supercomputers and big storage where we would be saved.

We still could die, for example somebody could destroy the server or a natural disaster destroys the server and we would be lost.

The server should live long enought that the human mind would evolve in some form or another that they won't need the server anymore.

So we need a way to make a supercomputer/server near indestructible.

Making it 100% indestructible would be the goal but I can't think of something really indestructible.

How do I make the server/virtual world indestructible?

Things to keep in mind:

• The goal is 10.000 years
• Natural disasters could happen all the time, so the server should survive these
• Where would we put the server?
• What protection will it get? (Just buried under the earth?)
• What components would be used to make it near indestructible?
• This world is set in a semi-far future, and we could place it anywhere in our solar system, so not just the earth

Answer 1

Redundancy You do not need to make a thing indestructible. You need to make the data undeletable. Imagine I wanted to destroy the word of God as written in the Bible. It would essentially be impossible. I could get rid of every bible in town but there are more in other towns. There are millions and millions of them. Even an apocalypse would miss the one in the forgotten nuclear bunker in North Dakota.

Have your database multiply redundant. Infinitely redundant sounds impossible but for a science fiction you could make it so - data coded in protonic spin, or the positions of gluons or something like that.

If you are super duper redundant then losing your data is less of a possibility. Corrupting your data becomes a bigger threat. Consider the data in our own DNA: multiply redundant with a copy in every cell. But when a copy goes bad and is not caught by repair mechanisms: cancer.

Has this circled back to Agent Smith?

ADDENDUM I was thinking about the ramifications of virtual life with infinite databases. Maybe all the databases are backup and only one is being used. A cool thing, though (to be revealed halfway through the story) is that they are actually all being used. There are infinite parallel virtual worlds with versions of us in each. This becomes like the parallel dimensions as seen in many shows (and my favorite episode of Trek:Next Generation) but you do not need scary spacetime rifts - it is extremely easy to hop from one reality to the next.

Answer 2

You don't make it indestructible.

You make it decentralized. You make it redundant. You make it run in something like the futuristic version of Google's server farms, where your search or your email or your doc hits some random server out of a pool of available servers. If one of those guys crashes, literally no one notices except some poor, overworked, engineer who has to swap the hardware and build a new server node in the farm. Then you have farms in multiple sites on multiple continents / planets / solar systems (scale out as needed for your specific Sci Fi vision).

Put it like this. Netflix has a "Chaos Monkey" that intentionally crashes their systems. (their name, not mine.) This is a utility that randomly kills things on their production servers. They do this because it forces their engineers to design systems that can withstand interruptions or crashes.

This way, even if a natural disaster DOES disrupt your entire building / city / continent / planet / solar system, no one dies.

You will still need network engineers -- even if they are AI-driven robots -- to go around, replacing dead parts. Maybe even software engineers replacing bad code.

But it's significantly cheaper to build commodity-grade hardware that can survive for a few years to a decade than to build an everything-proof hardened data center that's indestructible.

With sufficient scales of redundancy, you don't even need to really plan for major disasters. You just wait and when a center goes offline, you auto-fail to the backup site while recovery people/robots rebuild or repair the old site. Or if your disaster took the site off the map, you build a new offline center somewhere else and keep going.

You have two measures that matter.

In the plans for this, you basically have two measures that matter...

Recovery Point Objective (RPO)

RPO measures how much data you’d lose if your server went down and you brought up the secondary server. It’s measured in time, not megabytes. The less time you have, the more frequently you have to get the data to a secondary location.

Recovery Time Objective (RTO)

RTO is how much time you’ve got to get your server’s act together. Some companies measure it from the time you learn about the problem, some measure when the service is first unavailable. The clock’s starting point matters a lot.

In your scenario, both of these measures are basically zero. So you've got some work to do. And there's no way to build a single "box" strong enough to make these measures zero. So you have to have more than one box.

You still need an infrastructure and the industry to support that. So maybe there's a robot army out mining asteroids for raw materials to build new servers. Maybe that is controlled by your people or by a good AI that can do the work for your people.

Answer 3

10 000 years doesn't need to take 10 000 years

I'm not sure I fully understand this requirement:

The server should live long enought that the human mind would evolve in some form or another that they won't need the server anymore.

But if what you need is to spend 10 000 years on the server, just run the server really quickly. If you're running the simulation 1 000 000 times faster than reality (i.e. it can simulate 11 days of virtual time every second), you only need to keep it running for 4 days. And if something bad happens and the server fails, the meat-brained human is still alive so you can just run the thing again, no harm no foul.

Answer 4

I have one word for you: maintenance.

For something to be indestructible for that long it has to have a system of maintenance. And those have to have back up systems, as well as back ups of those back ups.

Just the BUILDING your server is in is going to need maintenance.

Stuff needs replacing, even in the future. Foundations crack.

So you will need nanites, and you will need robots or androids, or else it won't be pretty.

Robots will also need to be repaired and maintained.

Your server will be indestructible in that it will be maintained and protected. The idea of just leaving it be without that..you need to watch Life After People.

Your scale is 10,000. Even with advanced materials which do not yet exist on earth, destruction happens, unless you maintain.

So, what you will need is a series of failsafes in place. And you don't want to think single server here--over 10,000 years, that way lies death. Everyone knows you always have a back up...

1) General blue print for the whole system is present in all maintenance droids.

2) Droids should be in locations other than just one.

3) There's a planetary defense system for objects from space.

4) Multiple memory caches rather than just one. If one fails, you have back ups. There could be a relay system with old copies of the system. Scary bit is that people could lose time and not know it, because everyone has lost time--because the server in one location is destroyed or isolated, for one reason or another.

5) There will be mining for raw materials and a supply chain, because it will be needed. Earth will be easiest, even if they are bots, because we have a variety of materials, and conditions easily allow for construction and things like blow torches. Another planet's not out of the question, but Earth has too many advantages to ignore. (Although there are liabilities such as flooding that might not happen on a planet without water--but it will be a planet without water. And you'd be surprised how useful water will be in maintenance.)

The goal is 10.000 years a very long time.

Natural disasters could happen all the time, so the server should survive these You would need a complicated predictive modeling system in order to prepare for those via maintenance bots.

Where would we put the server? Underground is best. But one location is not a good idea.

What protection will it get? (Just buried under the earth?) Active defenders in the bots.

What components would be used to make it near indestructible? There's lots of bleeding edge tech in this area, carbon nanofiber, self-healing concrete are place to start, but microscopic maintenance might be a bigger help than the creation of materials.

This world is set in a semi-far future, and we could place it anywhere in our solar system, so not just the earth Earth will make things more and less difficult in ways I have outlined above. We are used to working in our own gravity and atmosphere, so it would certainly be easier to build an efficient system on our own planet for something this advanced.

Answer 5

You're going to want to work your way up to a Matrioshka Brain at some point. Since that's our goal anyway, let's go the whole hog.

Your server is going to be part of an interstellar spaceship. I'll assume that if you're able to upload brains, you have at least fusion power. Your ship's outer layers will be fuel tanks to hold hydrogen; below that will be a nice chunk of hull to keep everything inside where it belongs. Next layer will be raw materials, silicon, iron, oxygen, that sort of thing. Next layer, maintenance robots that keep everything running during the cruise. Finally, at the centre of the onion, you'll put your servers, safely held behind several miles of structure. Liquid hydrogen will cycle through the servers to keep them cool, then cycle back out to the outer hull to radiate that heat away into space.

Your ship will head for the nearest red dwarf star, Proxima Centauri, under fusion power. It'll take a couple of years to accelerate up to around ten percent the speed of light, then the engines will shut down and only a small onboard fusion reactor will keep running to maintain power to the servers.

During the flight, the robots will maintain the servers, make sure everything's running nicely. Out in interstellar space, everything is cold and quiet; no disasters to cause us any hassle.

The ship will cruise for fifty years or so, then decelerate towards Proxima Centauri.

The great thing about red dwarfs is that they're so very stable. They'll burn quietly for trillions of years without throwing a hissy fit the way our sun is likely to.

Once the ship reaches Proxima Centauri, two things will happen.

Robots will be sent out, with or without human minds to accompany them - your choice - and start to mine the system for materials.

Other robots will remain with the ship and start dismantling it, converting it into one or more static structures with solar collectors on one side, servers on the other. These platforms will be a home of humanity for the rest of time.

The materials-gathering robots will begin assembling two new starships, and more platforms. The starships will be loaded with new copies of humanity, and head out for more stars. The platforms will continue to multiply until all the material in the system has been used up, or the star is completely enclosed inside orbiting platforms - a Matrioshka Brain.

Answer 6

Several possibilites. Here are some ideas how to get there, but please be aware that you can only get very close to 100%. I do not want to talk about materials because this sounds like a dead end to me.

a) Do not localize your server. Spread out your storage. Check out "the internet"

b) Have someone or something repair the broken parts.

c) Put it into space or on a very boring planet/moon where there won't be many natural disasters. Take Pluto for example.

d) Make your advanced civilization advanced enough to be able to calculate for example i) the trajectory of every object in space that could destroy the "server" or network if you put it into space or ii) all natural disasters that will happen wherever you put it. iii) Anything else.

I could do that all day and at the end we might reach a number close to 100%.

Answer 7

Redundancy and replication

Redundancy has been covered by others, there's nothing i have to add here. But let's face it: nothing will last for 10'000 years and still function if it is more complicated than a brick.
So you need to replace things from time to time. This in turn means you need replacement parts. And unless you want to stockpile everything you might possibly need, and at least five times as much as you are going to need to account for failing infrastructure in the storage and whatnot, you should be able to create everything you need to replace every bit of your system.

This is where i actually have good news for you.
We are corrently well on the way of automatizing more or lesss everything, at an increasing pace. So as soon as you have autonomous robots that are not only able to provide maintenance to themselves and to automatic factories for (pretty much everything) but also able to re-create everything that they typically maintain, including themselves, you should be set. Now all you need is a multi-redundant system that will notice failing elements and be able to replace them, and your self-replicating maintenance robots that keep everything working smoothly.

The only thing your robots cannot produce themselves is energy. But conveniently, there is a very large fusion reactor nearby that won't stop providing "free" energy any time soon. So let's consider that settled, too. Because, really, humanity having advanced enough to be able to set up that system, it should be safe to assume they learned to stop digging up dead dnosaurs to make light...

Where do we put that beastie? Right here, on earth, spread around the globe.
First of all, this thing will have to be built. And some people will want to supervise the building. And being able to breathe while one does so is generally considered convenient. But this planet has more to offer: A magnetosphere to protect against radiation, and an atmosphere that burns up small objects moving at very high velocities. Both are commodities not to be sneezed at, so let's make use of them. The fact that you can make use of existing infrastructure also is a bonus.

Answer 8

Since there isn't any requirement for the server to be on Earth my thought is to send it out of the solar system into interstellar space. You could hand wave some type of 10,000 year power source powered by the radioactive decay of such and such material with an absurdly powerful and energy efficient super computer which should be coming soon if Moore's law holds for that much longer. Instead of interstellar space you could also just put it in far orbit with the sun in the Oort belt which allows you to use solar panels although then you are subject to things such as solar wind and the like. At that distance collisions are still astronomically unlikely in a 10,000 year time frame especially since scientists can accurately map and predict the trajectory of orbital objects on that timescale. In interstellar space there are 100 atoms per cubic meter which is essentially nothing for it to collide into. For comparison a single grain of sand contains 3 x 10^22 atoms.

Better yet create some redundancy and send off millions of these ships. Even if 99% of them fail due to some technical error a few thousand still exist and can achieve the next stage of human evolution.

Answer 9

Make it into a humongous mecha with weapons, and tools, and the ability to clone itself. In this way, if it foresees a catastrophe, it will be going to put a good use of the server's capacity and ask the humans "What to do?"

In this way, the server will have an active, and not just a passive defense, Planetary Annihilation sessions also tend to took so long, because the main unit, the commander can flee and rebuild the base elsewhere.

As far as components go: Graphene and other carbon allotropes, as it's strong, light (perfect for mechas), inert, a good conductor, and you can easily make it into a 3d material by making it into a gyroid shape at the molecular level:

Answer 10

Your question sounds very similar to one of the central plot lines of Greg Egan's Diaspora: it spends a lot of time discussing life in a polis (the hardware hosting a simulated reality for digital intelligences called citizens). Polises are located in isolated locations (e.g. deep under the Siberian tundra), and have multiple redundant copies (with merging protocols to allow citizens to combine the experience of their separated instances).

Answer 11

Others have already mentioned redundancy, replication, and maintenance, which I would combine to say: start with X redundant (and hardened) systems, and any time one is damaged or even threatened, build two more.

Answer 12

There is only one type of system known which can last this long, and that is a biological one which self-repairs. You could encode the data in some kind of ecological system, or something mimicking one. This could be a big soup of bacteria, or a swarm of nanobots or something like this. If computation speeds are an issue with this system, people could live out their lives at much slower than real time without them (i.e. 10 years for the stored people could be 100 years of 'real' time).

Watch out though, evolution always finds a way ...

Answer 13

Re visiting the idea of nanites. Lets say you had giant swarms of them that passively float along the ground or maybe swarm in the air when not needed. They break/die, but the other ones assemble new raw materials from the ground. The gen 2 will obvious upgrade themselves, and each generation will get better and better.

At this point, in the best case scenario they can actually convert atoms to other atoms.

Also plenty of redundancy in multiple locations.

So everything is humming along fine.

BOOM

Earthquake a server farm falls in and is destroyed.

The swarm comes in, and to as much degree as possible breaks the building into tiny specs, and move it a safe distance away. In the case all is lost, the swarm takes the elements from the ground, and makes a brand new server center atom by atom. There will be billions and billions of nanities so speed shouldn't be too big of an issue.

The server plants may either have nanities that complete build new parts and rescue materials from the old and/or have 3D printers.

Eventually replicators aka Star Trek will be invented.

Click boom, new server materializes from energy. Click and the old server is recycled back to energy. As long as the nanities can repair the replicators everything is good.

The key here is that robots and machinery are too prone to failure, where as nanities would self repair and regenerate.

Obviously each server farm would have to be self sufficient with solar,wind, water,diesel, gas, or whatever you got generators and backups.

It is best to give up on hard drives and go total SSD, solid state, storage. Solid state storage has too many advantages to ignore.

---

If you can't change atoms into other atoms, large wasteful production plants will have to be built. The nanities will then have to fly the raw materials that they mine all over the planet to product planets. Other will have to move those new items to where they are needed.

IF everything works properly server farms will randomly fall into lava pools, the oceans, and cracks in the planet, and they will simply be rebuilt a new elsewhere. Maybe you can make some air tight, and actually intentionally place them underwater.

---

The other option is of course space ships with the same general idea in mind. Nanities floating around everywhere doing maintenance. Some kind of space scoop to keep a steady supply of raw materials flowing. Even just in earth's orbit or near earth's orbit would probably get you 10,000 years easy. Moving again when our sun become unstable in whatever time that takes.

Answer 14

If virtual worlds exist, then they may be nested. Logic shows that the vast majority of worlds are simulations, in this case. So, the world containing the server is probably a simulation too.

Discover how to access and exploit that. Either move your own simulation “up” a level (gaining orders of magnitude performance boost!) or set an indestructible attribute on the server in the universe’s own simulation.

Answer 15

Simple. I don't know if you've read any Isaac Asimov but from a book in his Robot series, Robot Dreams, there is a computer that slowly evolved to be moved by humans into hyperspace, the upper dimension. Indestructable, it eventually became God to another species. It is assumed that the same thing would happen again, but I'm getting off-topic. My suggestion is to put it in hyperspace.

Answer 16

The coming paragraph doesn't answer your question per se so feel free to skip it: The concept of people living in a computer system is unachievable with our 1s & 0s. That being said, quantum computing could- in theory- be possible. Unfortunately it's in opinion that you could only transfer thoughts and memories; Desires and emotions wouldn't make the transfer, but hey, you would live "forever" and perhaps it could be simulated. The main controversy is people wonder if it is "you" that's in the system when you're transferred.

The system you're thinking of dissent nearly exist and as others have already said you need a redundant system. So far machines cannot run on their own there still need an "acolyte" of some sort either human- or other bio electrical being that heals technician the first one is more plausible. People that only purpose is to serve the system.

Even with all redundancies if the "server's" (or servers') environment in the real life world is destroyed Ott becomes in hospitable for it then the system could fail. To help counter act this a probe could be set to launch into doable if earth becomes inhospitable- even if it's a space station it would have to leave at a moment's notice. Then theirs a question if it will or not survive the transit to another planet even it assuming it's not looking for a habitable one. So in that case there should be more than one capsule splitting the load to limit collateral damage.when got think about it space would be the best option. You then have the problem of deleting memories of persons of a system loses data so the "persons" that do survive so that they don't feel lost or if the system resides to delete a few to save memory or power.

If your system doesn't use an "upload" of consciousness like the brain of the the persons' in a type of stasis would help remedy the allocation of memory space and ram it would also allow the system to tap on the physical brains as processing power for both the environment and problem solving. This would also help remedy the lack of emotions especially if the brains are properly maintained as well as fed the proper chemicals and hormones to achieve emotions in those acolytes I mentioned could be a brain placed into a bioelectrical android temporarily outside the "environment" l to perform the requires maintenance everyone would have to pitch in our be "expelled" permanently from the system.