Ask Slashdot: Protecting Data From a Carrington Event? 386
kactusotp writes "I run a small indie game company, and since source code is kind of our lifeblood, I'm pretty paranoid about backups. Every system has a local copy, servers run from a RAID 5 NAS, we have complete offsite backups, backup to keyrings/mobile phones, and cloud backups in other countries as well. With all the talk about solar flares and other such near-extinction events lately, I've been wondering: is it actually possible to store or protect data in such a way that if such an event occurred, data survives and is recoverable in a useful form? Optical and magnetic media would probably be rendered useless by a large enough solar flare, and storing source code/graphics in paper format would be impractical to recover, so Slashdot, short of building a Faraday cage 100 km below the surface of the Moon, how could you protect data to survive a modern day Carrington event?"
Don't panic! (Score:5, Informative)
First step is to stop listening to the hype. Yes it would be bad for the large power distribution infrastructure but no solar flare is going to erase optical discs that doesn't also wipe out most life on the planet. It isn't going to erase hard drives that aren't destroyed by the power events that happen in the first few minutes. So a copy in your safe will still be readable. Remember, the safe is metal and entirely enclosed. In other words it is a Faraday Cage. I really don't know how flash memory will react to a strong electro-magnetic field but my money on it also surviving so long as it isn't connected to anything when the balloon goes up. Kinda hard to induce much of a voltage across nanoscale features. And these observations also apply to an EMP attack.
It things really get bad you might have trouble finding a working system to connect that backup to and electricity to start it up with but if it gets that bad you won't be worrying about the source code to some damned game, you will be worried about God, Gold and Guns at that point.
While making those elaborate plans to protect your data you might also want to take a few precautions to ensure you are there to need that data when the dust settles. Do you have a bug out bag? Is it fresh? Do you have an escape plan? Odds are that if you are an indie game dev you live in one of the hives where venture capital can be found and everyone there is toast within days; the trucks stop rolling when the gas pumps stop working, the shelves empty and canibalism begins. Do you have a destination in mind? Do you have a few days of survival supplies stashed to allow you a chance to get to it?
Least of your worries (Score:5, Informative)
The only mechanism I can think of which would case a solar flare to render optical disks unreadable would be radiation damage. A solar flare which delivered that kind of dose would likely wipe out all life on earth so you probably wouldn't be worrying about your backups.
Yes (Score:5, Informative)
Punch Cards.
UPS Datacenter (Score:2, Informative)
I once toured one of THE TWO UPS datacenters that could run for a week on its own diesel generators. They figured that after a week, they could get more fuel to their generators if need be, but you also have to think that if things are so bad that critical infrastructure like the UPS operations cannot get power back after a week, then there is probably a disaster so big that the data might not mean much to anybody for much longer.
In the same way, if things are so bad with data drives and computers being destroyed everywhere in the world, who do you think is going to give a crap about being able to play your game?
Just a disk backup in your safe... (Score:4, Informative)
A steel box is a perfectly good Faraday cage. Its a small antenna cross section, so you'll effectively get no effects inside the box.
So if you are paranoid enough to care, just keep a backup of your data in your safe. Which you want to do anyway, since that helps mitigate many many many more risks to your data than a big solar storm.
Re:Don't panic! (Score:5, Informative)
You are correct about flash memory not being particularly vulnerable to EM. I work in data sterilization and modern degaussers that are used by a lot of government agencies to nuke their hard drives are completely ineffective on solid-state drives.
Missing sense of scale (Score:2, Informative)
The geomagnetic effects of a solar storm is very large, in terms of physical extent, but still relatively slow, small changes in terms of magnetic field amplitude. In a crude sense, you can think of the voltage induced in a loop of wire from a changing magnetic field. The voltage is proportional to the area, size of change in magnetic field, and how fast the field changes (inversely proportional to the time the change takes). The second and third factors are already quite small. The only reason it affects power grids, is due to their large size, allowing for a accumulating the effect of magnetic field changes over a wide area. This isn't going to affect small devices on the surface of the Earth, short of temporary power loss and possible loss of satellites from increased particle flux in space.
So how do you protect equipment like hard drives and computers from geomagnetic storms? Don't hook up antennas to them that are miles long. Or if you do plug them into the power grid, protection circuits are pretty simple for over voltage issues. So simple, a lot of power grid equipment has such circuit breakers, which is the reason they probably will go down from a major storm: not from damage, but from protection circuits pulling the plug when conditions go too far out of spec.
Re:Don't panic! (Score:5, Informative)
According to Wikipedia, a storm of this magnitude happens only once every 500 years or so.
Since one just happened about a hundred years back, the question is largely irrelevant.
While you're reading Wikipedia, look up "Gambler's fallacy". The fact that such an event occurred relatively recently has no effect on the probability that will happen in the near future.
Re:Optical? (Score:3, Informative)
How would optical be wiped by e/m radiation?
1. Find a CD or a DVD with your valuable source code on it.
2. Take the media to your kitchen. Insert media into microwave oven.
3. Turn microwave on exposing media to EM radiation.
Re:OK, I'll Bite (Score:1, Informative)
Kactus Games [kactusgames.com.au]. I really doubt the world would care if the source code to his not even released game was lost.
Re:let's see...linux kernel source (Score:4, Informative)
Or you could use Bradbury's method, get a bunch of people to commit it to memory by reciting it continuously.
Re:Don't panic! (Score:5, Informative)
aluminum substrate to write on
No. that's not right at all. The aluminum is only used as a reflective layer. A CDR/DVDR is:
1. printed label/printable surface
2. aluminum
3. dye
4. clear plastic substrate
On a blank disc the laser goes through the dye and is reflected by the aluminum.
When the laser writes to the disc, it (basically) burns the dye.
When the burnt area is read by the laser, it is not reflected back by the aluminum. (so now you have 1s and 0s)
Re:Last reason is a fail (Score:4, Informative)
Why would you assume that? CME's are related to sun spots and sun spots have definite cycles.
Re:How about something more plausible (Score:5, Informative)
I'll give a real world example of a time that a company I was doing work for got sued for about $8 million for license non-compliance. The company 15,000 seats and only had 5000 licenses of a product that they were using. The company demanded an audit and discovered that were 8000 installations of their software on site. Sound pretty open and shut, write a check, true up the licenses and somebody get's fired, right?
The data had to preserved /exactly/ as it was for the date of the audit. I performed both a back up of the database as well as a physical copy of the database. This was performed for an agreed upon date for the audit. I then preserved it on a separate database instance on a distinct SQL server owned by the legal department. I made sure that I did not 'clean up' or otherwise make the data look nice or presentable.
For those wanting to learn from this I was able to kill the lawsuit dead in it's tracks. When I talked with corporate legal counsel I asked them a simple question. Did the license say 'concurrent' or 'seat' in the fine print? Legal counsel came back to me an hour later to let me know that the license said "concurrent" and not seat. The fact that we had 8000 installations was meaningless /if/ I could prove that we never exceeded 5000 concurrent users.
I was able to do this by going back to multiple back up copies of the SQL database. I made copies of those database backups and put those on the legal teams SQL server. I then looked at the application metering data for the previous two years and was able to prove that we never exceeded 4700 'concurrent' licenses at a given time. This data was provided to their legal counsel in raw form.
To answer your question it boils down to this. You can keep your data as long as they can see your data (as necessarily redacted) in a /timely/ manner and without alteration. Speed is critical and you absolutely have to be able to show how you got your results and have to be able to replicate your work. In other words the other guys IT people have to take your raw data, write their own report and still get your results.
Both optical media and magnetic media are immune (Score:5, Informative)
Both optical media and magnetic media are essentially immune to solar flares. Hard drive electronics may be damaged, but the data will still be on the platters.
Magnetic tape is hard to erase; it takes a big magnet within inches of the tape. Degaussing most modern tape cartridges takes a field strength above 1000 gauss. The earth's magnetic field is around 0.5 gauss. It varies during solar flares and other events, but the numbers are all below 1 gauss. MRI scanners are in the 500 gauss range, and at those field strengths, metal objects become projectiles.
Magnetic tape is not affected by even intense gamma radiation. NIST totally settled that issue decades ago by lowering a recorded reel of 3/4" computer tape into the gamma ray pool of their nuclear reactor in Gaithersburg, MD, and leaving it there for 45 minutes. It then read back fine. Heat is a big threat to magnetic tape, though.
Re:Don't panic! (Score:2, Informative)
Not even remotely funny.
Just in case, from http://www.merriam-webster.com/medical/potable [merriam-webster.com] :
Definition of POTABLE: suitable for drinking
Re:Don't panic! (Score:4, Informative)
Re:Don't panic! (Score:5, Informative)
This is true of CDR's and other write-able media but NOT of (most) mass-printed optical media (though cheaper mass-printers have moved to this model now to save on industrial equipment).
Mass-printed media is much less likely to be damaged because they don't even contain dye. They are effectively PHYSICAL storage. A layer in the disk contains little bumps and dips. Dips reflect the lazer back, bumps scatter it, thus giving you 1's on dips and 0's on bumps.
This is one reason why you may well find your CD's from the 1990's still play fine even with a lot of scratches while CD's you bought in the mid-2000's are unusable. Many of the latter are dye-printed, which is much cheaper - but a far less reliable thing, the slightest discolouration throws the reader off while bump/dip optical media it's the actual physical shape of the disk that holds the information which is much less vulnerable to wear and tear.