Did Sunspots Crash Reddit on October 22, 2012?

On October 22, one of the sites where I publish came down with a sudden, bad case of glitches and load problems. While kvetching about the bugs with fellow writers, I started to blame "Sunspots!" by default. Then I began to think. Hey, wait a minute. Didn't I see on the news that Reddit and parts of Pinterest went down because Amazon Web Services had a server failure?

On a whim, I checked space weather sites to see if there WERE any active sunspots. Sure enough: kaboom! There's an active sunspot cluster which burped a ferocious X-flare on...October 22. Here's NASA's video of the flare in several wavelengths:

Luckily, the Earth's ionosphere — the ozone layer — and magnetic field absorb or deflect the brunt of solar flare energy, so we are safe from solar radiation on the ground (or in airplanes, by the way: they fly below the ozone layer and magnetic field).  

An exceptionally strong X-flare can cause radio blackouts, disrupting communications from police scanners to airline traffic control to your WiFi network. It can also throw GPS systems off by tens of meters, which is merely annoying when you're driving, but could be a problem for planes landing using GPS or big oil companies drilling in the arctic (here's why).

Just to give us fits, the radiation from a solar flare also chews on the ozone layer and causes the Earth's atmosphere to expand and drag down satellites faster than their orbits would normally decay. 

The real problem is those CMEs, the Coronal Mass Ejections. That's a chunk of the sun splashing against the Earth! No, it's not going to cook us, but if those charged particles hit Earth's magnetic field at just the right angle, a really powerful CME can cut through Earth's magnetic field and cause voltage spikes across the world's power grids, especially those closer to the poles or in New Jersey.

These voltage spikes are unlikely to damage your electronic devices permanently, but could cause short-term disruptions. The problem is, as Quebec discovered in 1989, that transformers can fuse, circuits can melt, and whole power grids can be taken out for significant lengths of time by very rare and exceptionally powerful solar storms. Your little smartphone isn't big enough to be hit by much solar energy, but a power grid can turn into a giant, multi-state collector for all the electromagnetic energy raining down from above. 

Luckily, power utilities are aware of the problem and can power down if they know a massive flare is coming. NASA now has a couple brave little satellites out there watching the sun 24/7, guarding us, although it's hard to predict which tiny percentage of all the flares they see will actually be hazardous to Earth. We need to learn more, so power companies don't get caught with their pants down. And we need to build more robustly-shielded electronics systems that can handle the occasional voltage spike from on high.

Researchers: The video below has links to official and important websites monitoring current sunspot/solar flare activity. Click video's title and check its description on YouTube.

The October 22 X-flare did not launch a massive solar booger at us, ahem, a Coronal Mass Ejection, which is the usual cause of voltage spikes on the ground. Earth's ozone layer probably got a good workout absorbing X-rays, gamma rays, and all that yummy solar radiation. 

Still, I'm intrigued about the timing. The NASA video above shows the solar flare peaked at 11:17 EST. TechCrunch's article covering the AWS server problems and outages at Reddit, Pinterest, Foursquare and elsewhere show the trouble started around 1:38 EST and got worse over the next few hours, while CBS reported that the disruption started around 2:03 EST. Did the sunspot continue to pump out enough jolts across the electromagnetic spectrum to cause some of that, or is it all just a coincidence? It's tempting to see a connection. 

Nevertheless, after sifting through science articles on solar flare impacts and tossing out all the doomsday conspiracy theory websites that believe that the Mayans could predict armageddon when they couldn't even predict their own civilization's demise (actually, they didn't predict armageddon, just a calendar reset, but that's another topic) — anyway, I suspect that glitches and outages for Amazon, Reddit, Pinterest and Squidoo on October 22 were due to good, old-fashioned, terrestrial gremlins. Apologies for the headline. 

The thing I find frustrating is that it's annoyingly difficult to find out just what the chances are that a big data server could be damaged or disrupted by electromagnetic pulses from solar flares. This is partly because most of our internet technology is young enough that we simply don't have years of data to draw upon, and partly because most of what's known is classified or hidden by companies trying to maintain industry secrets.

According to the science article I linked to above, small electronics and computers are liable to experience temporary glitches at most that can be solved with a reboot. Computers have a certain amount of shielding to protect them from voltage fluctuations, nearby lightning strikes, drunken college students sticking magnets on them, that sort of thing.

But what about big huge data centers with masses of computers hooked together that would dwarf even my college's 1989 VAX mainframe? They've got shielding, of course, but like power grid stations, they're a lot bigger, with more parts to fry. At least they aren't sending out massive power lines across hundreds of miles of open space acting as a ginormous spiderweb to catch whatever the sun throws at us.

Still, I'm not sure Google's data center stormtrooper provides sufficient protection against an 1859 Carrington Flare sized event. And once again, even if the servers are not damaged by solar flares, if power goes out for a large region, the web goes out, too. Then again, no Reddit is the least of our woes when, as this NASA study points out, the fridge and toilet aren't working. 

Out of random curiosity, I just went outside to see if I could spot the sunspot du jour with my wee backyard telescope plus the solar filter I jury-rigged for May's solar eclipse and transit of Venus. Yep, there it is: AR 1598 on the left side of the sun. 

If I've gotten myself oriented properly, flipping the image horizontally to compensate for my telescope's lens and mirror, it looks like AR 1598 is rotating towards us. It can take about ten days for a sunspot to rotate to the opposite edge. However, this isn't something to get too excited about, since there's been many X-flare capable sunspots for the last couple of years. Just like all the earthquakes in California each and every day, we're oblivious to sunspots unless, very rarely, they actually do something that affects us.

(The dark fuzz at top is from a passing cloud. Grr. Cloudy day in California, what's with that?)