Good morning!
Welcome to issue number 57 of Last Week in AWS.
This week brings me to Seattle for Microsoft Build until Wednesday. When I’m not live-tweeting the keynotes I’ll be wandering around like a lost puppy angrily trying to find a Starbucks that isn’t the one I’m standing inside. This week’s issue is sponsored by DigitalOcean:
Looking for a high-performance cloud platform with no billing surprises? Always know what you’ll pay each month on DigitalOcean with monthly caps and flat pricing across regions. Get started with a $100 credit. https://do.co/lastweek
Community Contributions
Event-based data synchronization is an underused pattern for data migrations; I’ve used it myself a couple of times over the course of my career.
The results of a Serverless survey are now available. Giant feature velocity increases, bill reductions, but be forewarned– this isn’t a change for the faint of heart.
A great deep dive into turning traditional tasks into Lambda– in this case, video encoding. Not only does it save money, but it’s a pattern that carries over very well to other similar tasks.
A fun tutorial about using the little-known - operator with the S3 command line utility to work with S3 objects directly via STDIN / STDOUT.
Tying Kinesis Firehose and osquery together for better MacOS desktop security is one of the more innovative pairings I’ve seen lately.
A quick tour through the strange and confusing service offerings that comprise the AWS IoTuniverse.
Someone at realtor.com got so irritated with ECS that they built “one that works” in CloudFormation. I can’t do it justice– go read A Better ECS.
AWS needs to hire Jerry Hargrove immediately; his hand-drawn AWS service icons have so much more soul than the official diagrams.
This time we’re talking about how Netflix optimized Flink for AWS. You should be more like Netflix, ignoring entirely the fact that you’re a multinational bank while they pretty much just stream movies.
This 10,000 word treatise on the fundamentals of RedShift absolutely blows me away. My knowledge of it more or less extends to “unless you want to pour money into it by the truckload, consider Athena instead.”
It never hurts to go back to basics– in this case, on a walking tour of the AWS Management Console. Enough of this changes often enough that it’s worth reviewing how it looks today, particularly if you’re not a frequent console user.
Last week I got to chat with Justin Brodley of Ellie Mae on Screaming in the Cloud. Check out Episode 8: A Corporate Prisoner’s Dilemma.
Tom McLaughlin of Serverless Ops crafted a well-reasoned think-piece on the “why” of Serverlessthat I have no real reason to disagree with other than to taunt Tom as is my tradition.
Choice Cuts From the AWS Blog
Access Amazon VPC resources for training and hosting with Amazon SageMaker – I’m going to pretend that it requires SageMaker models to figure out how the heck my VPCs are laid out. In reality, this is the first time AI has been known to cry and give up.
Alexa for Business now provides shared device connection status – Another new feature to “Alexa for Business,” or as you may know it, “the Great Satan of Open Plan Offices.” ALEXA, BUY MORE CHEETOS!
Amazon DynamoDB Global Tables Regional Expansion – It’s nice to see the asterisk next to “Global” getting at least a bit smaller.
Amazon S3 Adds Support for Amazon Glacier and S3 One Zone-Infrequent Access to Amazon CloudWatch Storage Metrics – You now get better insight into data you’ll never look at again, yet continue to pay for. At least now you’re paying less for it.
Announcing the General Availability of AWS IoT Analytics – You now get to know a bunch of data from your smart devices, like for exactly how long your bathroom scale has been attacking DNS servers.
Introducing Amazon EC2 Fleet – If you’re working in the finance department of a large AWS customer and you’ve painstakingly built out prediction models to optimize your usage of on-demand, spot, and reserved instances, you’re about to begin swearing like a sailor: EC2 Fleet now either renders that work obsolete or throws a giant question mark into the equation. I feel your pain.
Announcing Amazon EC2 H1 Instances Price Reduction – Let me save you a click. H1 are storage optimized instances launched late last year that few people have seen in the wild. Don’t feel bad; I had to look it up too.
AWS CodeBuild Now Supports Local Testing and Debugging – “You can now test your code locally” is a great feature enhancement announcement only until someone asks “wait– you mean you couldn’t before?”
AWS CodePipeline Supports Push Events from GitHub via Webhooks – Amazon admits that CodeCommit gets orders of magnitude fewer users than GitHub and bows to the inevitable of making the latter a first class citizen.
Welcome to “Ask AWS Abby” (or, congrats! it’s a blog!) | Dear DevOps Abby – A new blog has launched at AWS; AWS DevReloper Abby Fuller is now your “agony aunt.” She’s looking for questions to address in this weekly blog. I’ll get you started: Dear #askawsabby, My boss refuses to use cloud services, insisting we can build a better, more reliable platform for less money. Where can I find job listings for companies that aren’t run by lunatics?
Tools
This well-crafted IAM policy allows users to manage their own passwords and MFA devices– but nothing else unless they’ve authenticated with MFA.
Last week I open sourced one of the half-dozen Lambda functions involved in the production of this newsletter. It’ll require tweaking to work in other environments / use cases, but every hour I trigger a backup of my Pinboard account for later processing into the email you receive each Monday.
A graphical client for DynamoDB? Sign me up. You can pretty much think of “Dynamon” as phpMyAdmin for DynamoDB, which is a phrase that should cause endless screaming.
…and that’s what happened Last Week in AWS.
