WEBVTT

00:00:00.000 --> 00:00:05.760
Oh, look at that! Another major online outage that means you can't book travel, find a date,

00:00:05.760 --> 00:00:10.960
or order that refill of toilet paper you so desperately need, seriously send help. If it

00:00:10.960 --> 00:00:15.680
seems to you like many of the sites you depend on are going down more frequently, you're definitely

00:00:15.680 --> 00:00:20.720
not alone. But it's not like Facebook and Amazon are running on Windows Vista, so why the heck do

00:00:20.720 --> 00:00:26.640
these multi-billion dollar services seem so unreliable? It might be easy to blame some of the

00:00:26.640 --> 00:00:31.680
high-profile outages we've seen over the past few years on the COVID pandemic due to the huge

00:00:31.680 --> 00:00:36.800
amount of load placed on cloud servers from folks working from home or otherwise staying in. I mean,

00:00:37.440 --> 00:00:42.640
it's a lot of Zoom calls. But while stay-at-home orders and remote work did play a part,

00:00:42.640 --> 00:00:47.760
critical services experiencing downtime was already becoming an all-too-frequent problem

00:00:47.760 --> 00:00:52.960
before the pandemic. And it's still a big concern even now that much of the world is trying to get

00:00:52.960 --> 00:00:58.960
back to normal. But like most wide-scale problems, there are many things that play. For starters,

00:00:58.960 --> 00:01:05.040
brown-the-clock internet usage became very common very quickly. Think about the explosion in always

00:01:05.040 --> 00:01:10.720
connected IoT devices, people leaving high-def videos playing all night so they can fall asleep to

00:01:10.720 --> 00:01:16.160
them, and massive game and software downloads that often take place while users are in bed.

00:01:16.160 --> 00:01:20.880
These things just weren't that common only 10 years ago. But the explosion in net traffic since

00:01:20.880 --> 00:01:26.240
then has overloaded infrastructure that was only designed to handle peak traffic during certain

00:01:26.240 --> 00:01:31.520
hours of the day. But with all the resources that big tech has, why does this problem seem to be

00:01:31.520 --> 00:01:37.520
getting worse instead of getting better? So the fact that big tech is so, well, big is actually

00:01:37.520 --> 00:01:42.640
part of the issue here. Even though the internet is a pretty big place, there are actually a fairly

00:01:42.640 --> 00:01:49.040
small number of services that help run a huge number of sites. For example, Amazon Web Services

00:01:49.040 --> 00:01:54.400
provides backend cloud computing for everything from your Netflix recommendations to your Roomba

00:01:54.400 --> 00:02:00.560
vacuum to your Venmo account to even your airline reservations. And of course, AWS has had numerous

00:02:00.560 --> 00:02:06.960
outages that have made headlines with causes as diverse as server cooling failures, nasty weather,

00:02:06.960 --> 00:02:12.480
overloaded systems, and DDoS attacks. Amazon does have redundancies built in, but even so,

00:02:12.480 --> 00:02:18.720
the fact that one company helps support so much net traffic leaves one very obvious point of failure

00:02:18.720 --> 00:02:24.080
for lots of services that you didn't even know were reliant on Amazon, or one of its few large

00:02:24.080 --> 00:02:30.800
scale competitors like Microsoft Azure or Google Cloud. And to top it all off, your ISP is routing

00:02:30.800 --> 00:02:36.560
much of this traffic over cable lines that haven't received upgrades to improve capacity in a very

00:02:36.560 --> 00:02:42.560
long time, which can lead to congestion in the last mile of wiring that goes to your house.

00:02:42.560 --> 00:02:48.080
And there isn't even enough fiber on the backend right now to keep up with projected increases in

00:02:48.080 --> 00:02:54.800
data usage, especially as mobile carriers continue to roll out 5G. So until cloud systems become more

00:02:54.800 --> 00:03:00.240
secure and more redundant and telecom providers build out better infrastructure, I would not

00:03:00.240 --> 00:03:06.160
count on 100% uptime. But the good news is that these outages give you an excuse to stop mindlessly

00:03:06.160 --> 00:03:10.640
scrolling through your phone and go touch some grass. Thanks for watching guys. If you liked this

00:03:10.640 --> 00:03:15.440
video, hit like, hit subscribe, and hit us up in the comment section with your ideas of topics that

00:03:15.440 --> 00:03:17.040
we should cover in the future.