Thread by @QuinnyPig

Hello, and welcome to Day 2 of the #AWSSummit in San Francisco. @SwamiSivasubram's keynote starts in a few minutes, and I'm trying something a bit different for today's livetweet thread....

I'm here today *strictly* as a customer. Through a careful strategy of "declining meeting invitations" and "outright ignoring people," I know absolutely nothing about what's in store for us today.

I'm not even allowed to sit in the press/analyst session!

The building is nowhere near capacity, which given... well, everything, makes sense. I exercised good judgement and left my "AMI HAS THREE SYLLABLES" protest sign at home.

It feels like that's not the cultural mood at this moment.

The keynote starts with a video about how closely @intel and @awscloud are working together.

Any bets on whether the keynote starts with AWS talking about how awesome their Arm-based Graviton processors are?

The smoke machine really elevates the sadboi feel of the energy so far.

Ooh, Databases are now under Swami?! I have a new person to ask about Route 53 issues!

If you want to learn some deep magic, bias for the chalk talks. They arent recorded and people let things slip they otherwise wouldnt.

The Original Original Home of EC2 is almost certainly some sad-sack wiring closet with power issues.

Ooh they updated the @CloudNativeFdn landscape diagram with new graphics!

AWS trumpets its relentless focus on innovating, mostly around new databases and ways to run containers.

Occasionally, such as in the case of RDS Custom, both.

And yet if you ask an @awscloud person how to deploy something like lasttweetinaws.com to every commercial region simultaneously, they look at you a bit wild-eyed and back away slowly.

I wonder how formally I would have to draw up the 17 ways to run containers on AWS in order to get it cited in an @awscloud keynote.

Predicts that way more workloads will move to the cloud in the years to come.

I find that boring, honestly. I want to see the new workloads that will only be built on the cloud; moving your data center basically rounds to an AWS cash grab.

As someone who sorts through every last one of these for inclusion in @LastWeekinAWS, I think @awscloud owes me an apology.

I guess @awscloud is now planning a region on the moon or whatnot.

Ah, the old "solid fuel boosters' size is dictated by how wide an ancient Roman road was, dictated by two horses" analogy.

The irony of AWS pointing out "high technology is determined by how big a couple of asses have been" is not lost on me.

AWS is continually departing from the past. For instance, it's pretty clearly not Day One anymore.

Nitro is AWS's own custom hypervisor. It unlocks a lot of really neat capabilities, and a few insufferable conference slides.

"We have over 500 instance types, more than any other cloud provider!"

Other, more sensible cloud providers: "Y'know what, @awscloud? You can have that. Knock yourself out."

Customers: "Oh god."

I will never, ever forgive @awscloud and @intel for not debuting the Habana accelerators with a parody of Barry Manilow's "Copacabana."

Here we go. Amazon Graviton Processors. THIS KEYNOTE HAS BEEN SPONSORED BY @INTEL

Swami continues to give a survey about the breadth of AWS services. If you've been following me for a while, you've heard about most of them already. If you haven't been, you're most likely fairly happy; I don't want to ruin that for you.

And now an AWS testimonial from @tacobell. This is the weirdest sponsorship deal I've seen so far this year.

@googlecloud preps its Next conference, sponsored by @Wendys.

This is exactly my point.

Swami: "Cloud is for every application!"

Data egress starts at 9¢ a gig. I promise you, @awscloud is absolutely not for every application.

Talking about how awesome @awscloud's global network is. For the price, it damned well better be.

Okay, something in the iPad Pro's color correction makes this slide... less than helpful. I assure you, it makes sense in person.

There were probably better weeks for @awscloud to drop a Netflix reference.

AWS Outposts, like giraffes, don't really exist. Or if they do, I can't get one without Enterprise Support and a loading dock. Like giraffes.

"But enough about Netflix, let's talk about NASDAQ." Oof, how quickly the knives come out.

A manufacturer of fiber optic cables talks about how they're using @awscloud services to make the cables, which they then sell to AWS.

This is basically an episode of "How It's Made" for your AWS bill.

"VP of Edge" is forevermore known as "AWS EdgeLord."

This is a survey of AWS shit-tier service names.

The re:Invent preview of IoT TwinMaker is now Generally Available.

Good for the customers it helps, but that's very far out of my wheelhose.

It lets you conduct tests of changes without turning one of your warehouses into the Stanford Prison Experiment.

Amazon: "Yeah, we're still gonna do that second thing anyway."

I wanted to like Amplify Studio but what I really want is "It to build the backend for whatever I do in Figma." It's just not built that way, or at least in a way that resonates with me.

Anyway, it's generally available today.

Now the GM from Mapbox comes to the stage. It takes him a few minutes to get there because Location Services are *screwy* around Moscone.

I have to wonder if Mapbox would have built on AWS if Amazon Location Services had been released before they got started.

And Swami's back to gloat. Or so I assume. "Modern apps store MASSIVE amounts of data!" hits differently when it comes from a company that charges you for it by the gigabyte-month.

"Statistic provided by Forrester."

Months ago at @forrester: "Hey, what's your AWS bill trajectory been like?"

AWS has been selling AI/ML for an awfully long time into this market for this statistic to not be deeply unsettling to some customers.

Swami: "Job 1 has to be security!"
@awscloud: "Security is job zero."

IAM: "Off by one error! Oh no!" *fails open globally*

Yeah, we need some more databases.

It gets confusing, y'know?

"What makes this database different than all other databases?"

--The Four Questions of AWS Passover

Amazon Aurora Serverless V2 is now GA after a lengthy beta period.

And Serverless V1 basically wasn't.

Here's the @awscloud storage portfolio. These are databases with really complicated query methods.

"Data Movement Services" is very much "money printing machine go brrrrrrr"

AWS: "Tools for your data lake!"
Customers: "We don't have one of those."
Me, the AWS Whisperer: "They mean all the shit you've got hanging out in S3."

Swami: "When building a data lake, there's no better place to start than S3!"

See? I told you.

AWS Glue is handy, until you start to do too much with it.

"AWS Glue Autoscaling is now generally available."

Yup, that's doing too much with it.

GA for AWS Glue Sensitive Data Detection.

This is going to lead to a SageMaker-style cambrian explosion of branded Glue features, isn't it.

"Customers have told us how easy it is to get started with Athena." They're right! It is!

As one of those customers, I can also report how easy it is to migrate to @SnowflakeDB once you get serious about your data analysis.

Now announcing Athena data source connectors, including Snowflake. Well, that would have been interesting if it worked the other way. And was released a month ago.

Now a Salesforce customer reference. They're speedrunning all of my new vendors apparently...

Murali Krishnaprasad (@salesforce EVP of Engineering) is talking about the real-world value of combining disparate data sources.

I usually find the topic boring, but he's just a *riveting* speaker. No joke.

Holy crap, an architecture diagram that doesn't expose all of the complexity to get its point across. And the slide design is on point.

Amazon's "we're not a PowerPoint company" begins to present as an area of weakness.

My colleague Jeremy joined me today to watch me livetweet this keynote. As @jbrodley once said, "it's like watching a sarcastic unicorn give birth."

Having used both:

Amazon QuickSight is for Business Intelligence.
@tableau is for Business Wisdom.

I hope this works out better than S3's one-click public access granting from a decade ago.

Ooh, the QuickSight product roadmap preview!

And now an @F1 video. "There's HUGE amounts of data coming off of the vehicles!"

@awscloud: "And we can bill on *ALL* of those dimensions!"

Absolutely everyone else: *backs away slowly*

SageMaker Serverless Inference is generally available today. We *almost* made it out without another SageMaker offering...

Now you can use Textract Queries to extract insights from documents using plain language.

And now a new guest on stage. She comes out 3 minutes after when the keynote was supposed to conclude. Amazon Time Sync is apparently drifting again.

Some things are designed to be easy / effective to operate.
Other things are designed to look good in live demos.

The @awscloud console is neither of those things, which is why it's odd to see it demo'd in a keynote. They usually don't do this.

While yes, Machine Learning® is becoming accessible to all customers, I'd posit that it's not delivering value for even a majority of them.

"Teach 29 million people to use our cloud services" is one approach, but I'm partial to "make those services accessible and intuitive enough that many of those 29 million people don't need extensive training to unlock value."

Now another Deep Racer promo. I'm tempted to sponsor a team.

Swami closes with a shout-out to "Learn and Be Curious."

And the keynote is over! If you've enjoyed this, @LastWeekinAWS has similar content / humor every week in the newsletter.