Welcome to the tenth issue of Last Week in AWS.
Last week was relatively slow in the world of AWS; no ground-breaking feature releases came out, just a handful of solid improvements to existing offerings. It’s telling that AWS advances so quickly that the lack of an earth-shaking announcement for seven days is itself worthy of comment.
Community Contributions
If you’ve got time to read a deep-dive PDF, check out theAbstract for Bolt: I Know What You Did Last Summer… In The Cloud. It’s a great dive into what you can tell about other guests on the same physical EC2 host in AWS.
Check out the story of Moving One of Capital One’s Largest Customer-Facing Apps to AWS; it’s a great example of the sorts of things large companies consider when doing a cloud migration. There’s a bit of lesson and tooling in here for everyone.
How AWS Cloud is demolishing the cult of youth is making the rounds, discussing how Amazon is somewhat rare in the world of tech with respect to how it tends to hire people with a few decades of experience.
A Comparison of Advanced, Modern Cloud Databases — Brandur Leach – A terrific comparison of the modern “cloud” databases. This is a fascinating capability survey of Aurora, Spanner, Cosmos, and several others.
Ticketea Engineering – Scaling Amazon Aurora at Ticketea – Autoscaling Aurora is an interesting challenge. Ticketea discusses how they scale Aurora up and back down again for preplanned load spikes.
Choice Cuts From the AWS Blog
Build a Serverless Architecture to Analyze Amazon CloudFront Access Logs Using AWS Lambda, Amazon Athena, and Amazon Kinesis Analytics– If you’re used to grepping information from text logs, or perhaps massaging them with awk, get with the times; proper log analysis for CloudFront requires tying together S3, Kinesis, DynamoDB, Lambda, S3 some more, Kinesis Analytics (which apparently is a completely separate service from Kinesis), and no less than $100 a month to get up and running. Interesting concept, but generally overkill for many use cases. Looking forward to the upcoming Part II, which is slated to include visualization / another half dozen AWS services.
Amazon QuickSight now supports Federated Single Sign-On using SAML 2.0 – You may be thrilled that another service now provides “federated single sign-on using SAML 2.0.” If you have no idea what any of those words mean, oh how I envy you.
New Features for IAM Policy Summaries – Resource Summaries – See what resources are affected by your various IAM policies. If you’ve grown tired of the tried and true IAM diagnosis method of guess-and-check-and-scream-in-frustration, this one’s for you.
Amazon Elastic Block Store (Amazon EBS) Now Supports Cost Allocation for Snapshots – EBS snapshots have been an exception to cost allocation tags; you can assign them tags, but they won’t work to assign cost / blame. That changed this week, as a poorly understood billing area became a bit more transparent.
AWS Knowledge Center Video: Preparing to Send a Snowball Back to AWS – Jeff Bar (AWS’s chief evangelist) gives a 90 second video rundown on how to package a Snowball for return to Amazon. This video gives a good sense of a Snowball’s physical scale, as well as highlights why AWS Snowball has replaced “a ball of packed ice” as the snowball with which I would least like to be hit in the face.
Tools
GeoEngineer from Coinbase brings power to the people who said “I love Terraform, but what would be amazing is wrapping it in a ruby-like DSL.” Snark aside, it’s worth a look.
Eric Hammond’s ec2-consistent-snapshot attempts to flush and freeze the filesystem (along with a flush and lock of the database if applicable) before taking a snapshot, in an attempt to improve snapshot consistency. Very handy– but pay close attention to the Caveats section.
Tip of the Week
Data transfer in AWS can be very expensive. It’s also difficult to plan for; if you don’t believe me, take a look at how data transfer billing works. I’m working on a new version of this graphic, but despite being very deep into the data transfer costing weeds, I still reference this image no fewer than three times a week.
…and that’s what happened Last Week in AWS.
