Welcome to the third issue of Last Week in AWS.
The Sydney AWS Summit ended Friday, and the San Francisco AWS Summit is scheduled for April 18 and 19th. I’ll be there asking a variety of uncomfortable questions to AWS employees; if you’ve got a burning question you’d like me to ambush them with, hit reply and let me know.
VPC concepts are sometimes tricky to wrap your head around. Not only does AWS VPC Core Concepts in an Analogy and Guide explain them in an approachable way, it also includes a great image of the AWS documentation as a people-eating bear.
Last week’s coverage of Medium’s move to the cloud focused on the architectural aspects. They’re back this week with a blow-by-blow of the cutover itself, in Cloudy, With a Chance of Meetup: Behind the Scenes of Moving Meetup to the Cloud.
There’s been a lot of noise about AWS Organizations over the past few months. Exploring AWS Organizations gives a great overview of the service in ways a human being can understand.
AWS and Canonical have teamed up to release a high performing Ubuntu kernel for Ubuntu Cloud Images. Better IO performance, faster boot times, better networking and more now work automatically on any official Ubuntu 16.04 AMI released after March 29th. If you’re using Ubuntu in your AWS environment, this is well worth checking out.
Choice Cuts From the AWS Blog
Coming in 2018 – New AWS Region in Sweden – Sweden will host what’s scheduled to be the 19th AWS region. A boon to Europe’s patchwork quilt of data privacy laws, we’re looking forward to seeing what eu-north-34 looks like.
Amazon Elasticsearch Service now supports up to 100 nodes and 150 TB storage capacity per domain – while it’s nice to see managed Elasticsearch able to scale beyond where it historically could, I’m hard pressed to identify too many shops who work at this scale and aren’t already managing Elasticsearch themselves.
New – Host-Based Routing Support for AWS Application Load Balancers – This handy feature expands the routing features of ALBs, while simultaneously making their billing model even more inscrutable. ALBs are billed based upon new connections per second, active connections per minute, bandwidth passing through the load balancer, and now rule evaluations (number of rules – 10, then multiplied by your rate of requests). This dramatically simplifies the answer to “What will our ALBs cost us?” down to the very honest and brief “I couldn’t tell you if my life depended on it.”
In the world of AMI creation, Packer 1.0 was released last week. Widely regarded as the definitive tool for AMI (and other platform specific images!) creation, it’s hard to identify a better method to repeatably create custom AMIs.
Truffle Hog is a great way to scan entire git repositories (including history and branches) for high entropy strings, and an insulting thing to call your dinner companion when they snag the last appetizer. In real world terms– it helps you figure out if someone committed a secret key / password to a git repository before you release it to the world. I’m tempted to fashion this into a git hook.
Tip of the Week
It’s nerve-wracking to pull the trigger on a big Reserved Instance purchase for a variety of reasons. One of the common reasons that keep people from doing it is the concern that upcoming AWS price drops may make their purchase ill-timed.
If AWS reduces pricing that impacts a purchase you’ve made in the past three months, stories abound from multiple customers about Amazon “making things right” with a service credit. While this is neither advertised nor guaranteed, it’s never been in Amazon’s interest to make their customers regret giving them money. Despite the snark I throw their way every week, I’ve never known them to treat their customers unfairly.
…and that’s what happened Last Week in AWS.