3 AWS Talking Points We Stole

Posted by Jen

4 December 2014

Did you know that Weather.com is more accurate than AccuWeather? Neither did we. This was just one of the fun tidbits Mingle team members Kiera and Rao learned this month from the 2014 Amazon Web Services re:Invent conference in Las Vegas.

The AWS product launches and announcements have been covered well elsewhere, so we will just say that we were anticipatory fans of AWS Lambda and that we met a cool data migration company called Cloudly that seems like it could save a lot of headaches. A lot of trends like Big Data and microservices were covered at AWS too, but here are our three top talking points we heard.

1) “Yay for failure!” Netflix had a great session on how failure is inevitable but if you plan well, you can have it happen when most of your users are busy elsewhere. Netflix’s resilient system is nearly stateless (except for caches) so if requests fail, they can switch to another server. As Netflix puts it, everything should be redundant (and isolated). Maybe if HBO GO took this attitude, it wouldn’t crash during “Game of Thrones”.

2) “Find someone who’ll give you honest feedback.” A good mentor can be a key to success, especially for women in the industry. This sentiment was from the Women in Tech luncheon, which was so popular with attendees of both genders that it ran out of chairs. Kind of ironic that some women were not able to attend, especially due to the low number of women at Re:Invent, but it’s a great sign that industry players are recognizing sexism as a company problem. Maybe next year AWS will have more female speakers overall and plan for higher attendance.


3) “Think ahead and learn from others, even if you don’t have 2 billion customers.” Best example of this was probably the dreaded “Bart screen” that players of “Simpsons: Tapped Out” encountered repeatedly due to EA’s terrible experience with outsourced development that eventually required the company to move development and re-launch in 2012. One specific problem (among many) was database inefficiency due to an oversized cluster. A solution (conversion to DynamoDB) took three months. Learn from EA, Netflix, MLB: Don’t be the Bart screen.

And with that, Skrillex says it’s Peanut Butter Jelly Time.

comments powered by Disqus