The outage in Amazon Web Services (AWS) popular storage service S3 a couple of days ago was severe. Over 50 businesses who entrusted their websites, photos, videos and documents in S3 buckets found themselves unreachable for around 4 hours. Among those were high profile names such as Disney, Target and Nike. And it’s not the first one either. This time, again, the outage took place at Amazon’s veteran Northern Virginia (US-EAST-1) region.
Amazon’s own websites, however, were not affected by the outage. According to Business Insider the reason is that
They have designed their sites to spread themselves across multiple Amazon geographic zones, so if a problem crops up in one zone, it doesn’t hurt them.
Put simply: Amazon designed its websites the right way – with high availability and disaster recovery plan (DRP) in mind.
If you want your website to sustain such outages – follow Amazon’s example! Here’s a piece of advice I wrote a few years ago after another major AWS outage:
For more best practices on resilient cloud-based architecture check this out:
And if policies, regulations or your own paranoia level prohibit putting all your eggs in Amazon’s bucket, then you may be interested in this:
So Keep Calm – there is Disaster Recovery!
Follow Horovits on Twitter!