Explore T5 Datacenters
Request a Tour

5 Things We Can Learn From The AWS Outage

According to SimilarTech 148,213 companies rely on AWS S3. 3-4 trillion pieces of data are stored on it. When there’s an outage, we all are impacted to some extent. As more and more companies adopt a cloud-first or cloud-only strategy, incidents like the one on Feb. 28th will become more impactful.

  • The Cloud is not a silver bullet. While many people believe using the cloud solves many IT-related issues, the fundamentals of computing still remain. Servers fail. Storage arrays fail. Applications fail. Networks fail. Whether your IT environment is in the cloud or not, any, and possibly all of these will occur. While you may not have to manage infrastructure in a data center, the infrastructure is still out there somewhere. In this particular case, it is believed to be an application issue that caused the outage. S3’s health dashboard was also down because it lived on S3. AWS drank its own Kool-Aid.
  • Service Level Agreement’s are important. We may never know the true cost of the 4 hour outage on Tuesday afternoon. Some companies may not get compensated for the outage. Having a good Service Level Agreement is your insurance policy when you put your company’s applications in the trust of others. Make sure you understand exactly what happens when those applications go offline. What kind of notifications will you receive? What amount of compensation will your company be entitled to? When can you expect to get back online?
  • Downtime is costly. A 2016 report by the Ponemon Institute estimated that downtime costs the average business $9000/minute. The AWS outage lasted 254 minutes. Using my iPhone calculator, those 148,213 companies could have lost an equivalent of $338B dollars yesterday. The average company could have lost over $2M yesterday. Granted, many had work arounds, but you can see that it can get expensive quickly.
  • Having a Plan B or Plan C is probably a good idea. Lewis Bamboo, an Oakman, Alabama small, family-run business had a contingency plan. When the outage occurred, it took all the images off their website. Luckily, they had those images stored locally and were able to redirect the image server to their local server, so customers could continue to see what type of bamboo plants were available for sale.
  • A Hybrid IT or Bi-Modal IT environment makes sense. Ask any Risk Manager about risk avoidance. They’ll tell you to spread out vulnerability to minimize the risk. It’s no different with an IT department that aligns the business processes of a company with the internal or external applications and systems of the company. Having some applications on-site might make sense. Having some systems in a colocated environment might make sense. Having an off-premises disaster recovery site might make sense. You spread out your risk and minimize the impact of an outage.

Bottom line: The cloud is an effective way to use consumable infrastructure service models to optimize workloads and create better alignment with the demands of your business. However, it should be used as part of an overall IT plan.



Written by Bob Kramlich

Bob is the Director of Enterprise Services for T5 in Atlanta. His primary role is to help T5 customers optimize workloads and create better alignment with the demands of their businesses.

Email Bob

Ready to talk about your mission critical data center needs?

We’re excited to talk about the data center business and what we can do to help you. We can start the conversation right now and schedule a tour today.
Request a Tour
Copyright © 2018 T5 Data Centers. Designed by Gumas.