14 ms·

I've always been hearing that solar flares and the resulting geomagnetic storms pose danger to hard drives and electronics. Is there any truth to it? Recently

• joshspankit 3y ago
  You can mitigate most of it by using ECC memory, a good line-filtering UPS, and some type of error-checking for reads/writes/at-rest. The last of the risk can be functionally eliminated by your off-site backup.
  • egeozcan 3y ago
    > off-site backup Underground or at the other side of the planet or both?
    • shagie 3y ago
      Not powered, but the next room over is acceptable for a solar flare. The issue would be that wires start getting induced current in them ( https://en.wikipedia.org/wiki/Geomagnetically_induced_current https://en.wikipedia.org/wiki/Geomagnetically_induced_curren... ) that arcs and shorts out the device. However, if you've got a backup tape there's nothing from the current that will flip the bits on the tape itself. There are good reasons to have the backup be off site (other physical calamities) - but the aurora isn't going to flip any bits on a backup. At worst, it will destroy the machinery from a surge from the grid and an extremely powerful one may destroy the machinery from arcing in the device itself (this would be extremely powerful)... but the media itself is fine.
    • joshspankit 3y ago
      I suspect you’re trying to be sarcastic and/or facetious, but: Both ideally sure. The more backups and more locations the better. Diminishing returns pretty quickly but why not if the data is important. Electromagnetic energy from the sun only has a chance of damaging equipment and only some of the bits of that equipment will be affected. You could have two servers in the same datacenter where one gets hit and the other does not. Or even two drives in the same chassis. And even then, it might not be all the data on that drive. Of course no security is absolute (learned that early from my uncle that owned a convenience store at a lake: He spent extra on doors, locks, windows, etc and someone drove a truck through the brick wall to rob the place). Same here: if a solar event happens that would wipe out all of your onsite and offsite servers “on the surface” then there is a much bigger problem.
      • egeozcan 3y ago
        > I suspect you’re trying to be sarcastic and/or facetious It really was an honest question. Thanks for the detailed response even though that wasn't obvious.
  • kaivi 3y ago
    > by using ECC memory ECC protects from bit flips, but aren't high-energy particles required for flipping a bit? AFAIK nothing in the solar wind interaction with our atmosphere can produce them. They'd likely arrive earlier than the wind, but I also haven't heard of gamma ray bursts to accompany solar flares...
    • shagie 3y ago
      They are most noticeable for x-rays, though there is an increase in gamma rays too. https://www.swpc.noaa.gov/products/goes-x-ray-flux https://www.swpc.noaa.gov/products/goes-x-ray-flux https://www.epa.gov/radtown/radiation-solar-activity https://www.epa.gov/radtown/radiation-solar-activity > Solar flares are large eruptions of energy coming off the Sun containing several different forms of energy: heat, magnetic energy, and ionizing radiation. The ionizing radiation released during solar flares includes x-rays and gamma rays. These rays of ionizing radiation can damage satellites because they are in space and are not protected by the Earth’s atmosphere. Magnetic energy from solar flares can interrupt radio communication on Earth or damage communications satellites. Unless you've got a satellite, this likely isn't impacting you. Bit flips tend to be from from heavy charged particles. https://www.scienceabc.com/innovation/what-are-bit-flips-and-how-are-spacecraft-protected-from-them.html https://www.scienceabc.com/innovation/what-are-bit-flips-and... I believe you're more likely to have a bit flip from radioactive decay in the chip itself than a cosmic ray.
• ceejayoz 3y ago
  Day-to-day ones aren't really worth worrying about. Something like https://en.wikipedia.org/wiki/Carrington_Event https://en.wikipedia.org/wiki/Carrington_Event would probably make quite a mess, of the nature where individual server rooms are probably a secondary consideration.
  • Lapalux 3y ago
    Wow! Telegraphs Because of the geomagnetically induced current from the electromagnetic field, telegraph systems all over Europe and North America failed, in some cases giving their operators electric shocks.[22] Telegraph pylons threw sparks.[23] Some operators were able to continue to send and receive messages despite having disconnected their power supplies.[24][25] The following conversation occurred between two operators of the American telegraph line between Boston, Massachusetts, and Portland, Maine, on the night of 2 September 1859 and reported in the Boston Evening Traveler: Boston operator (to Portland operator): "Please cut off your battery [power source] entirely for fifteen minutes." Portland operator: "Will do so. It is now disconnected." Boston: "Mine is disconnected, and we are working with the auroral current. How do you receive my writing?" Portland: "Better than with our batteries on. – Current comes and goes gradually." Boston: "My current is very strong at times, and we can work better without the batteries, as the aurora seems to neutralize and augment our batteries alternately, making current too strong at times for our relay magnets. Suppose we work without batteries while we are affected by this trouble." Portland: "Very well. Shall I go ahead with business?" Boston: "Yes. Go ahead." The conversation was carried on for around two hours using no battery power at all and working solely with the current induced by the aurora, the first time on record that more than a word or two was transmitted in such manner.[26]
    • post-it 3y ago
      It's neat that both telegraph operators seemed to immediately know what was going on and why. Professionals with good debugging skills.
      • mensetmanusman 3y ago
• f5ve 3y ago
  I can't answer your question directly. But a while back I did study up on the 100-year solar flare (i.e. the equivalent of the 100-year flood). I was curious exactly how bad this would be for society. As I recall it would be bad and do a lot of damage in the short term. Lots of things, especially in communications, would obviously be damaged and very expensive to fix. But hard drives would mostly be ok and therefore the most crucial pillars of society would emerge intact.
  • cout 3y ago
    What about SSDs?
• dallyo 3y ago
  Maybe look at material about EMP weapon effects, I imagine it's probably the same kind of risks. I think solar flare risk is overblown though.
  • crashingintoyou 3y ago
    > I think solar flare risk is overblown though. What you'll find on the insurance side of thing is this sort of claim from Lloyds of London (https://assets.lloyds.com/assets/pdf-solar-storm-risk-to-the-north-american-electric-grid/1/pdf-Solar-Storm-Risk-to-the-North-American-Electric-Grid.pdf https://assets.lloyds.com/assets/pdf-solar-storm-risk-to-the...): > A Carrington-level, extreme geomagnetic storm is almost inevitable in the future. While the probability of an extreme storm occurring is relatively low at any given time, it is almost inevitable that one will occur eventually. Historical auroral records suggest a return period of 50 years for Quebec-level storms and 150 years for very extreme storms, such as the Carrington Event that occurred 154 years ago. If anything I think the solar flare risk is understated.
    • captainbland 3y ago
      > 150 years for very extreme storms, such as the Carrington Event that occurred 154 years ago. Wow. I knew this was a possibility but I didn't know we were overdue. Echoes of COVID a bit as a "once in a hundred years pandemic". Looking forward to the space mandated holiday to be honest.
      • __MatrixMan__ 3y ago
        We have a lot more wires lying around than they did in 1859, so that's a lot more induced current than they had to deal with. They probably had all of the fires out within a day or two--I'm not sure we'd be so lucky. I wonder if we have enough replacement parts on hand to recover after an event like that. Given our labyrinthine supply chains, if it becomes a manufacturing bottleneck it could take years to retool. In a morbid way though, I'm also looking forward to the holiday. I'm really curious to see what changes we make while the lights and cameras and payment systems are offline.
• mannykannot 3y ago
  I'm nowhere near an expert, but the only mechanism that I have heard of for them causing damage is through inducing large currents in long power transmission lines (and, formerly, long telegraph and telephone lines.) I would guess that the risk to electronic devices themselves is less than from a nearby thunderstorm, and that measures to protect from damage via lightning's effects on your external power supply would also be adequate protection from geomagnetic storms.
  • bob1029 3y ago
    This is my understanding as well. The length/scale of the conductive element is directly proportional to how much danger is involved. The modern power grid has lots of places where it can trip nearly instantaneously upon adverse conditions. Assuming we can automatically trip any segments of transmission, I think most of the damage could be isolated. Long conductors that aren't connected to anything can't cause any damage beyond the lines themselves. I don't think transformers and generators sitting in storage are going to feel much of anything.
• rrock 3y ago
  Bit flips from cosmic rays. Earth’s magnetic field helps to shield us, but is disrupted now. I work with a CCD camera in a sub basement lab. On a normal day we see a stray cosmic ray every couple of minutes (as a hot pixel). On a day like today, might be every couple of seconds.
  • tudorw 3y ago
    annoying, but kind of awesome :)
  • cshimmin 3y ago
    True, but the hot pixels you see in the CCD isn't from bit flips. Rather the CCD is properly doing its job of collecting electrons. It's just that spurious electrons are being produced by the cosmic rays' ionization trail (as opposed to photoelectron in normal operation).
    • anamexis 3y ago
      What's the difference? Aren't bit flips also just spurious electrons being produced by the cosmic rays' ionization trail?
      • akiselev 3y ago
        It’s an excitation of the sensor. It may be interpreted somewhat like a bit flip when converted to an image, but it’s normal for image sensors to experience a lot of noise, going back to the days of film. It may just be eliminated by the denoising algorithm like most spurious electrons/photons.
      • godelski 3y ago
        Bitflip is when a high energy particle changes the state of a memory address or signal in a wire. Hot pixels are when the CCD (camera sensor) achieves its max level and becomes bright. Pixel overloaded (should also result in spillover). Difference is really just what part of the system receives the energy. For example, you can overload a pixel with a laser, but you also should get spillover. Bit flips tend to be more localized, affecting only one part. In reality, the difference doesn't matter too much, as long as we're keeping our discussion to cosmic events. But also note that images experience a lot of (non-gaussian) noise and plenty of this is from cosmic excitation as well as from other natural and man made sources. Just the levels are a lot lower and unlikely to result in a bitflip (which is a pretty high energy event).
• AuthorizedCust 3y ago
  I’m concerned that the repetitive, sensationalized news coverage will encourage apathy. Again and again, headlines blare doom narratives about flares, but I am unaware of major impacts recently.
• korse 3y ago
  You haven't missed too much. The cosmic ray bit-flip guys have a point, but that is very hard to shield against and probably won't kill your servers. See this paper for a deeper dive. https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=1562&context=spacegrant https://digitalcommons.usu.edu/cgi/viewcontent.cgi?article=1... In regards to the Faraday cage (grounded mesh), assuming you have a very good path to ground and a sufficiently robust mesh material, you are shielding against any emissions with a wavelength larger than λ. You will have better results versus wavelengths that are much larger however. If you want a deeper dive on that, you can actually read Faraday, or Wikipedia or just ask your search engine about a basic dipole antennae and work up from there. Many computer components are somewhat shielded/tolerant of standard levels of RF interference in the environment so you might not even need to bother (depending on application). Normally (in my experience) you see problems arise with sensors, digital control lines and some human interface devices in very noisy environments (industrial facilities with a lot of high power variable speed electric motors). Regarding aperture size λ, figure out what source of RF you're most concerned about, determine the range of wavelengths emitted, concentrate on the most powerful bands and then make your aperture smaller than that. Examples are 'the sun', cosmic rays, north korean nukes, the welding robots on the other side of the wall etc. Finally, if you're going to shield, don't half ass it or you waste money. Gaps around doors, or anywhere you poke cable through a wall can cause problems if not handled correctly.
• evgeniysharapov 3y ago
  Here's famous story about bit-flip changing/affecting results of election in Belgium https://en.wikipedia.org/wiki/Electronic_voting_in_Belgium https://en.wikipedia.org/wiki/Electronic_voting_in_Belgium
• marcosdumay 3y ago
  > What wavelengths am I protecting against? Something around hundreds of meters. It's large enough that people don't treat it like a wave. Also, it's often semi-stationary, so you have to care to make your protection long-lived, and not only surge-based. Coronal ejecta is formed by charged particles. The ones that can move only a few meters into the atmosphere. You don't have to catch those. There is also no high-frequency EM radiation.
• godelski 3y ago
  > Recently I've been building a server room, and naturally a question of EM insulation came up. If I were to protect my servers from this, I needed to wrap the room in a grounded mesh with holes of λ in diameter. What wavelengths am I protecting against? I used to build space radiation shields. I'll give you three answers. What you should do: Throw up a few sheets of metalized mylar or some aluminum foil. Make sure everything (including the foil) is electrically isolated and grounded. Be more worried about the grid more than the air. Don't let the grid kill your components, because it WILL be overloaded. You want to be very secure: Do the above, but be in a concrete building. Better if under ground. Idk, add some chicken-wire or other mesh? Don't ground to the grid, ground to the ground (use a grounding rod and bury it deep). The overkill (an explanation): Ground based radiation shields tend to be different than space based as you don't care about weight on earth. How to stop radiation? Mass. Even still, the two follow a pattern: layers of differing material. Plastics (sometimes doped with boron, titanium, or other materials) are good insulation and protect well against neutrons (not much of a worry for you unless we're talking nuclear weapons), but you'll still want some because these are layered with conductors (copper or aluminum). Conductors pick up charged particles (the majority of radiation). There is a lot of confusion about radiation and it took me awhile to actually learn this. People will talk about attenuation distances and {alpha,beta,neutron}-cross-sections, but if we're talking about high energy particles, then a lot of this doesn't actually matter. Mass does. So don't worry about optimizing materials or any of that, just put mass in the way. Reading told me one thing but testing and simulation told me another thing. These cross-sections and attenuation depths are mostly for lower energy particles (like you'd find in a reactor). Your mass doesn't have to include any metal, but that will help because the biggest component is that you'll want some grounding. Also, use ECC memory and other such server grade components.