Welcome to issue number 45 of Last Week in AWS.
Last week I asked for reports on folks using SageMaker. The response was overwhelming. It’s clear that AWS is onto something fascinating here.
In “coming soon” news, I’ll be speaking at Southern California Area Linux Expo; use “LAST” for a 40% discount on tickets to the conference.
I’ll also be keynoting DevOps Days Charlotte at the end of this month.
I don’t question the math in this S3 vs On-Premises analysis, but I’ve never yet found a comparison that was a true apples-to-apples S3 comparison. I’ve done client work contrasting the two, and it’s deceptively simple-looking from the outside; it’s a very hard nut to crack.
Eric Hammond has a post about how to spit out the various AWS accounts within an organization, in a format a human being might be able to actually do something with. You’re going to want to tuck this one away for future reference.
As container workloads proliferate, getting useful metrics out of them becomes a larger problem. This post talks through the process of distributed tracing with Istio in an AWS environment.
Cloudonaut goes on a deep dive into monitoring Amazon Elasticsearch.
OpsGenie has a great post explaining how to attach useful details to links expanded in Slack. It’s powered by Lambda, API Gateway and Node.
I can’t find a more public announcement than this AWS Forum post, but there’s a new CPU credit allocation policy for T2 instances.
This post talks through the process of getting daily DynamoDB backups via Lambda.
“Lambda Local Environments are Complicated” reports area man Ryan McCue, master of understatement.
A wonderfully written and researched postmortem on building deep learning on Lambda. Economics, performance, code… it’s all here.
AWSgeek is at it again, this time with an in-depth look into the magic of DynamoDB.
The EFF weighs in on the cutesy-named CLOUD act, which would enable cross-border cooperation for law enforcement to gain access to cloud data. This likely surprises no-one, but the EFF is not a fan.
The S3 Bucket Negligence Award returns, this time afflicting a brand marketing firm called Octoly.
If the idea putting an application into containers / Fargate feels like you’re sitting in the foothills of Mount Hopeless, you’re absolutely not alone. It’s a daunting task. This tutorialtakes you through the process step by step, and should remain valid for at least 20 minutes until they change the console UI again…
A gorgeously drawn comic book contrasting AWS vs. Azure is so bizarre to me that I was convinced that someone was pulling my leg– but it exists, and it’s glorious!
You generally don’t see companies coming out with reports touting their own impending demise, but that didn’t seem to stop Cisco last week.
Choice Cuts From the AWS Blog
10x Higher API Call Rates for Amazon Kinesis Client Library (KCL) Applications – You can now shove ten times the data into Kinesis, after which you will never look at it again.
Amazon Cognito Simplifies User Migration – Cognito takes some getting used to when you’re using it for toy projects, as I do. It’s a whole other kettle of pain when you’re trying to shove hundreds of thousands of existing users into it. This enhancement helps with that.
Amazon DynamoDB Now Supports Server-Side Encryption at Rest – On the one hand, this is a huge win for using DynamoDB in compliance-heavy environments. On the other, is “someone breaks into an AZ and steals drives out of the server racks” a realistic threat model in 2018?
Amazon ECS Adds New Endpoint to Access Task Metrics and Metadata – AWS lurches ever closer to being able to answer the question “what does each containerized task in our cluster actually cost us to run” without having to hire someone like me.
Amazon RDS and AWS Database Migration Service Support Replication from SQL Server – DMS becomes more capable all the time. At this point it’s largely become something you can trust; unfortunately the service was launched before it got to this point, so I fear that its proponents are swimming against a tide of previously-valid criticism…
Amazon Route 53 Auto Naming Announces Support for CNAME Record Type and Alias to ELB – Chalk up another loss for the “stop using CNAMEs” pedants; Route53’s autonaming now supports this much maligned/misused record type.
Announcing Responses Capability in Amazon Lex and SSML Support in Text Response – Amazon “No, Alexa is something different” Lex gains additional features to make chat bots suck less. I’m interested in this space, but struggle to find a good idea to build myself.
Introducing the AWS Instance Scheduler – “Click button, receive automation” that automatically turns instances on and off at various times of day. This has huge potential costing wins– imagine turning off your developer environments at night, on the weekends, during holidays, and whenever Apple has a keynote.
Voice Your Web Content with Amazon Polly Plugin for WordPress – As riveting as “Alexa reads Last Week in AWS to you” would likely be, I’m going to instead be launching my own podcast myself. More on that front in the coming weeks.
The Amazon WorkMail API Now Supports Exchange Web Services (EWS) Push Notifications – Workmail and Exchange now work better together, but I hope I never have to use either of them myself. I have strong opinions about email, closely held.
New AWS Developer Training in Collaboration with edX.org | AWS News Blog – I’m hugely excited about this. It helps give a canonical answer to the age-old question of “I set up an AWS account and what I thought was a bunch of fine print is actually their list of services. Oh please help what do I do?!” Anything that makes for a more approachable onramp to AWS is a big thumbs up in my book.
Watchmen centralizes management of various AWS Config rules across multiple accounts. If you’re using Config, check this out. If you’re not using Config, take this time to get caught up on your expense reports.
…and that’s what happened Last Week in AWS.