Give me an @awscloud service, product, or team and I will say something nice about them in this thread if it kills me.
Braket shows that not everyone at AWS is focused on "today" or "next quarter."
Chime is a terrific example of "willing to be misunderstood for long periods of time." It's not what you think it is. More on this to come when I have the time to write it up.
Amplify makes AWS infrastructure accessible to folks who neither want to nor need to care about its depths. That's no small thing.
Snowmobile is proof that that ridiculous question you dare not ask maybe should get asked anyway.
EC2: For all its might and power, the cloud is really just layers of abstraction atop virtual computers. Which are themselves another abstraction layer, but that wrecks the metaphor. It's a utility--you don't even have to think about it most days.
Cloud9 demonstrates that even if they think your use case is slightly nutty, there are teams at AWS willing to go all in on supporting it.
IAM serves as the wall against the darkness that keeps our data safe--but we owe it to everyone to meet them halfway.
Activate shows that there are people in this world capable of listening to some of these startups describe what they do while maintaining a straight face. And then helping them with their infrastructure needs.
Managed Blockchain shows us the dangers of listening to "customers" who clearly don't have signing authority. They further demonstrate that no matter what whacky shit you're into, @awscloud won't shame you for it.
AWSECS4K8S(EKS) showed us that AWS will listen to reason about names that are patently ridiculous, and fix them when pressed.
AWS IQ is fantastically useful--I say this as a paying customer.
"Upwork for AWS without a lot of Upwork's terrible problems."
"Upwork for AWS without a lot of Upwork's terrible problems."
SimpleDB shows that even when the CEO calls a service "a failed product" in a press interview, you won't be abandoned if you chose to build your business on top of it.
Ground Station shows that absolutely no customer will be able to use *EVERY* AWS service.
Even @jeffbarr had to pretend to have a satellite in his blog post about it.
Even @jeffbarr had to pretend to have a satellite in his blog post about it.
"Build an exabyte scale data processing system" is hard enough. "That nobody will ever appreciate and complain about endlessly" means it's an underappreciated technical marvel despite its flaws.
It's helpful to remember that just having someone click on the traffic lights is sometimes the better business outcome than a $40 million machine learning project.
It's helpful of AWS to allow us to respond to some tool saying "hurr durr the cloud is just someone else's computer" by dropping one of them on their foot.
It's sometimes nice to be reassured that our calls really *ARE* important to Amazon as we please continue to hold.
Even AWS clearly understands that the one true universal API is and always will be email.
Complain all we want about rough edges, we'd be the victims of a lot more data breaches without Cognito.
"We built a fantastically powerful exabyte scale query system, so all you have to do now is write an interface to query it; easy!" Friday afternoon requests happen even at large impressive companies.
"Take this code and run it" is really what a lot of customer requests boil down to. The early implementations got a lot more right than they did wrong.
S3 remains the eighth wonder of the world. Infinite storage (and it is infinite; fight me!) for pennies a gig.
Democratizing access to infrastructure is the real revelation of the cloud. We're starting to see the same thing with ML.
Monitoring is of the devil, and AWS didn't opt out. Recent changes indicate that they're really listening to the pain points from actual customers, not just Twitter randos.
"Take a look at all of this nonsense we built. Got it? Good, now explain all of it to people." It's like writing an encyclopedia about a world that refuses to hold still.
"Serve this content from over a hundred locations around the world, basically for free" is an incredibly tall ask that we all take basically for granted.
I won't give you a 100% SLA for love or money, but Route 53 will.
"All of the things that manage your environment inside the instances" used to be the realm of dodgy third parties. AWS is stepping up in a major way into a thankless niche.
HTTP is weird, and API Gateway supports all of that weirdness, for which it is then unfairly blamed.
"We just billed a customer for 20,000 distinct items. Please display this to them in a format that doesn't render them senseless."
"I want to get this message from here to there" has a dozen possibilities, but basically all of them build on top of these two services.
In theory, everyone uses infrastructure as code. In practice absolutely everybody needs this, but they lie about it.
Google's MapReduce practical joke years ago was so good that AWS was a good enough sport not to ruin it for people.
Any company will build a tool that lets you shovel your data into their offering. AWS built one that also works in the opposite direction.
In this time of quarantine, this team ensures the handful of travelers get to see a splash of color and a familiar logo.
"Run this one application in a region but pipe the output to my desk" shouldn't require a full workstation. And it doesn't.
Besides its actual business value, ELB acts as a great demonstration of abstraction. It does a lot under the hood that you can sorta feel working, but you don't have to *care* about it; it just works.
AWS offers hope that one day we can stop paying the ancient enterprise pricing models of Tableau.
QLDB is proof that AWS will not only talk to Blockchain afficianados, but make them actually define their needs rather than just babble on about blockchain.
Useful security data in emerging environments can in fact be detected automatically provided you are oh so very careful with where you point it.
"Working with data you've shoved into the appropriate place, formatted specifically" has one glaring flaw that Data Pipeline helps address admirably.
CloudFormation has saved us from GitHub repositories filled with 40MB of screenshots explaining how to set things up.
"We've seen an awful lot of code failures, let us spot them in your code in a secure, scalable way" is uplifting.
The @awscloud Solutions Architects are living proof that the wisdom to spot horrifying bullshit design patterns and the diplomacy not to say that phrase out loud when you do can indeed coexist within the same person.
Archimedes once said "Give me a lever and a place to stand and I can move the world."
HackerNews once said "I can fix that in a Python script over a weekend," and AWS Glue gives them a place to run it.
HackerNews once said "I can fix that in a Python script over a weekend," and AWS Glue gives them a place to run it.
"Running a script isn't hard; my watch can do it" is true until you try to run millions concurrently while managing no infrastructure and fielding questions from jackasses, all for pennies. Lambda delivers.
The future sounds nutty until someone builds it from building blocks like AppSync.
"Never roll your own crypto unless" and then there's a laundry list of nigh-impossible checkboxes that KMS checks.
AWS once used the phrase "push a button" to someone who'd never seen a computer interface before. To their credit, they owned the misunderstanding.
Without data egress, AWS would just absorb all of your data and never give any of it back. That's not a cloud provider, that's Facebook.
Amazing capabilities provided by systems that predate cloud? AWS has an answer for those applications, too.
Because you're not going to implement the S3 SDK in COBOL any time soon.
Because you're not going to implement the S3 SDK in COBOL any time soon.
I would like an S3 SDK in COBOL. #awswishlist
Lake Formation gives you a fighting chance of building an actual data lake instead of a data swamp.
"I want a VPS" isn't the kind of ask that should result in having to go to Cloud School for three months. AWS ensured it wasn't.
Migration Hub makes the "holy SHIT we forgot about the mainframe and there's no AWS/400" realization happen early enough in the process that it can be dealt with.
Lumberyard has one of the most amazing easter eggs in its terms of service of all time.
Rekognition single-handedly made that XKCD comic about "tell me if it's a picture of a bird / it'll take five years" obsolete.
STS has saved us from the nightmare of "emailing IAM credentials back and forth between companies."
Control Tower has spared you the horrors of watching me build a bad version of such a thing myself and posting it on GitHub.
Outposts shows that AWS is willing to listen to customers, even when the concern is "but I'm scared of the cloud and also the dark."
Amazon MQ demonstrates that interoperating with a terrible protocol doesn't require that you implement the terribleness under the hood.
Pinpoint ties together a bunch of disparate services in a productized way rather than "here's an AWS Solution that makes you do it all yourself." It shows an empathy that we'd all benefit from.
If you want to run actual-ElasticSearch but not pay for data transfer, Amazon ElasticSearch has you covered.
Turning the AWS billing dimensions into something for which "simple" isn't an immediate complete joke is the stuff of legend; this calculator delivered. I use it constantly.
I joke about using Route 53 as a database. Service Catalog makes it way less likely that one of your divisions actually tries to build something like that.
AWS Config is one of a small handful of services without which compliance with many things would be out of the question in AWS.
Not every developer should have to learn CloudFormation. With CDK, they don't have to.
A rare glimpse under the hood, Hyperplane enables some awesome stuff--and lets us peek at the complexity required to do it ourselves.
