Security Cloud Privacy Tech

Peter DeSantis’ Keynote at AWS re:Invent 2021

AWS re:Invent is always THE event in cloud. Petere DeSantis’ keynote provided a peek behind the curtain of the technology that drive AWS itself.

This is the Twitter thread of my coverage of the keynote…

Tweet 1/104  Next tweet

rolling now with the keynote from Peter DeSantis, SVP Utility Computing and Apps, @awscloud

#reinvent

Tweet 2/104  Next tweet  Start

this used to be Monday or Tuesday Night Live and is always one of my favourite talks of the week

#reinvent

Tweet 3/104  Next tweet  Start

Peter giving his perspective on the start of the @awscloud

#reinvent

Tweet 4/104  Next tweet  Start

big focus from the team before the launch was on the key word “Elastic”

<< I’m willing to be it still is!

#reinvent

Tweet 5/104  Next tweet  Start

early memories from the original EC2 team. 7/10 of the team are still @awscloud

#reinvent

Tweet 6/104  Next tweet  Start

email from @JeffBezos to the EC2 team in the early days. love the focus on the user experience

#reinvent

Tweet 7/104  Next tweet  Start

key areas of focus for EVERYTHING @awscloud

#reinvent

Tweet 8/104  Next tweet  Start

looks like we’re starting the night by looking at storage

#reinvent

Tweet 9/104  Next tweet  Start

remember, this is (usually) not a feature/service launch keynote. it’s typically a look behind the scenes @awscloud. lots of amazing technology and engineering challenges

#reinvent

Tweet 10/104  Next tweet  Start

remember when @awscloud looked like this?

#reinvent

Tweet 11/104  Next tweet  Start

lots of stuff build out from S3

#reinvent

Tweet 12/104  Next tweet  Start

for a longer look back at @awscloud’s early days. be sure to check out @jeffbarr’s session, “15 years of AWS with Jeff Barr”

https://virtual.reinvent.awsevents.com/login/

#reinvent

Tweet 13/104  Next tweet  Start

we’re going right down to the fundamentals. looking at old school(ish) hard drives

#reinvent

Tweet 14/104  Next tweet  Start

lots of mechanical engineering packed into these types of drives. Peter used a crazy airplane analogy that I won’t even try to summarize

lots of improvements but the mechanics are pretty steady

#reinvent

Tweet 15/104  Next tweet  Start

“Hard drives remain the best way to store large amounts of data when you need to access it immediately”, Peter << he left out, “…even though we don’t want that to be the case”

#reinvent

Tweet 16/104  Next tweet  Start

“Let’s do some quick math”, Peter << a key indicator as to why this is always my favourite keynote

#reinvent

(sorry @werner)

Tweet 17/104  Next tweet  Start

here’s what the math says : hard drives suck for these type of big data workloads

#reinvent

Tweet 18/104  Next tweet  Start

the opposite view for a large but low usage workload

#reinvent

Tweet 19/104  Next tweet  Start

basically the individual workload doesn’t align to the hard drive mechanics. you need to aggregate these workloads in order to smooth things out

#reinvent

Tweet 20/104  Next tweet  Start

Peter’s currently walking through how @awscloud distributes customer S3 workloads across the physical backend of the service

it’s the only way the math adds up

#reinvent

Tweet 21/104  Next tweet  Start

“If you want to innovate at scale, you need to move quickly. AND you need to do it safely”, Peter DeSantis, @awscloud

#reinvent

Tweet 22/104  Next tweet  Start

talking about how @awscloud approaches testing S3 now

#reinvent

Tweet 23/104  Next tweet  Start

I love this type of deep dive. especially the contrast in the complexity behind the scenes of S3 for something as simple as a bucket/key #ux

#reinvent

Tweet 24/104  Next tweet  Start

for testing, @awscloud is leveraging automated reasoning more and more. it’s the only way to truly test things out

more at https://www.amazon.science/

#reinvent

Tweet 25/104  Next tweet  Start

excellent timing by the @AmazonScience team, “A gentle introduction to automated reasoning” << just published

https://www.amazon.science/blog/a-gentle-introduction-to-automated-reasoning

#reinvent

Tweet 26/104  Next tweet  Start

formal tools are great. they have a very high bar for correctness

…but at the cost of speed and reducing the available talent pool

#reinvent

Tweet 27/104  Next tweet  Start

this isn’t a great choice to face. at @awscloud’s scale, they deal with it all of the time

…so they developed a new approach, “Lightweight Formal Methods”

#reinvent

Tweet 28/104  Next tweet  Start

here’s a great related paper from @AmazonScience , “Using lightweight formal methods to validate a key-value storage node in Amazon S3”

https://www.amazon.science/publications/using-lightweight-formal-methods-to-validate-a-key-value-storage-node-in-amazon-s3

#reinvent

Tweet 29/104  Next tweet  Start

lots of advantages to this approach…biggest challenge would be the initial design to allow for the application of these formal methods when required

#reinvent

Tweet 30/104  Next tweet  Start

ohhh, I beat Peter to the punch

#reinvent

Tweet 31/104  Next tweet  Start

first customer story up, this one’s from @Adobe

#reinvent

Tweet 32/104  Next tweet  Start

. @adobe’s intro covered how they are massive scale in the cloud, not turning to storage

#reinvent

Tweet 33/104  Next tweet  Start

talking about photos and how we stored physical photos

#reinvent

Tweet 34/104  Next tweet  Start

digital is a similar challenge but at a larger scale

how many photos & videos do you have?

#reinvent

Tweet 35/104  Next tweet  Start

next year “we” (not sure who that is, all of us?), we’ll generate 1,500,000,000,000 photos

#reinvent

Tweet 36/104  Next tweet  Start

. @adobe turning to the two primary storage use cases for storing photos and videos

“I want it now"

“I want it sometime, whenever really”

#reinvent

Tweet 37/104  Next tweet  Start

the obvious questions about this story form @adobe is why do I only get 100GB of storage with Creative Cloud then?

<< feature request == let me use my own @awscloud account of Creative Cloud storage & sharing

#reinvent

Tweet 38/104  Next tweet  Start

s/of/for/

Tweet 39/104  Next tweet  Start

learn more about @adobe Sensei at https://www.adobe.com/ca/sensei.html

#reinvent

Tweet 40/104  Next tweet  Start

. @adobe is obviously the reference customer for Amazon S3 Glacier Instance Retrieval

more on the @awscloud feature at https://aws.amazon.com/about-aws/whats-new/2021/11/amazon-s3-glacier-instant-retrieval-storage-class/

#reinvent

Tweet 41/104  Next tweet  Start

it snuck under the radar for most people but it could be quite a game changer

#reinvent

Tweet 42/104  Next tweet  Start

Peter moving on to block storage types now

#reinvent

Tweet 43/104  Next tweet  Start

now diving into SSD technology

#reinvent

Tweet 44/104  Next tweet  Start

SSD remove most of the mechanical challenges but they have their own issues…like the limitations of flash storage

#reinvent

Tweet 45/104  Next tweet  Start

primary the paging system & the lifetime of those pages

#reinvent

Tweet 46/104  Next tweet  Start

as Peter is delivering this section of the talk, I realize I know way too much of the low level functionality of both HDD and SDD.

this is the downside of nerd life. lots of obscure knowledge

…but I wouldn’t have it any other way

#reinvent

Tweet 47/104  Next tweet  Start

of course there are different issues at hyper scale that @awscloud has discovered with low level SSD tech

#reinvent

Tweet 48/104  Next tweet  Start

small inconsistencies add up fast at hyper scale. things we would never notice even with hundreds of systems, they see regularly

#reinvent

Tweet 49/104  Next tweet  Start

. @awscloud Nitro smooths a lot of this out

more on Nitro at https://aws.amazon.com/ec2/nitro/

#reinvent

Tweet 50/104  Next tweet  Start

…of course they built their own Nitro SSD

#reinvent

Tweet 51/104  Next tweet  Start

btw, @QuinnyPig it live tweeting this keynote as well. for his take ( ) head over to https://twitter.com/QuinnyPig/status/1466181160099139586

#reinvent

Tweet 52/104  Next tweet  Start

Peter citing the performance delivered by the custom SSD approach via @awscloud Nitro

#reinvent

Tweet 53/104  Next tweet  Start

Tweet 54/104  Next tweet  Start

on to @awscloud Graviton now…

#reinvent

Tweet 55/104  Next tweet  Start

more on Graviton at https://aws.amazon.com/ec2/graviton/

#reinvent

Tweet 56/104  Next tweet  Start

. @awslcoud is pushing Graviton everywhere. and why not? more bang for your buck in most cases w/cloud native designs

#reinvent


Tweet 57/104  Next tweet  Start

Peter touches on the Graviton3 update

#reinvent

Tweet 58/104  Next tweet  Start

have we seen these stats for Graviton3 yet?

…not that we should focus on them according to Peter (and I agree)

#reinvent

Tweet 59/104  Next tweet  Start

I like Apple’s performance per watt approach. which @awscloud has followed as well

#reinvent

Tweet 60/104  Next tweet  Start

more on Dennard scaling at https://en.wikipedia.org/wiki/Dennard_scaling

#reinvent

Tweet 61/104  Next tweet  Start

Peter talking about power consumption now. it’s a critical attribute for all of us, mind boggling at @awscloud’s hyper scale

#reinvent

Tweet 62/104  Next tweet  Start

How do we efficiently increase performance of a Graviton core?

Make the core WIDER

#reinvent

Tweet 63/104  Next tweet  Start

there have been lots of improvements in silicon design over the past decade. the easy way to think about it is for year all we worried about was the size of the engine

we finally realized that the tires, chassis, fuel, and other attributes matter as well…

#reinvent

Tweet 64/104  Next tweet  Start

…that means more cores, wider memory lanes, direct access to storage, etc.

this is why we’ve seen massive boosts in performance in things like @apple’s M1 and @awscloud Graviton3

#reinvent

Tweet 65/104  Next tweet  Start

another customer story up now, this time it’s Fannie Mae

#reinvent

Tweet 66/104  Next tweet  Start

Kimberly Johnson, EVP & COO of @FannieMae given the scale and scope of their operations

tl:dr = it’s big

#reinvent

Tweet 67/104  Next tweet  Start

ooohhh, @FannieMae built out an HPC solution @awscloud Lambda. would love to get a deep dive on that one…

#reinvent

Tweet 68/104  Next tweet  Start

very interesting solution from @fanniemae. they leverage some key @awscloud data services to expand credit score analysis to include rent payments…regardless of how they were made

#reinvent

Tweet 69/104  Next tweet  Start

now talking about evaluating risk to housing due to climate change. again another massive data problem, taking unstructured data into structure data for analysis and prediction

#reinvent

Tweet 70/104  Next tweet  Start

good customer story from @fanniemae. I think they struck the right depth to solve solutions to unique problems at a scale that’s only really possible in the cloud

#reinvent

Tweet 71/104  Next tweet  Start

on to #ml now. I swear this is slide is even more packed than @SwamiSivasubram from earlier today

#reinvent

Tweet 72/104  Next tweet  Start

Peter’s breaking #ml down into two big stages:
  • training
  • inference

you need different infrastructure & tools for both

#reinvent



Tweet 73/104  Next tweet  Start

diving into inference now…

#reinvent

Tweet 74/104  Next tweet  Start

more on @awscloud’s custom silicon, Inferentia, for #ml inference at https://aws.amazon.com/machine-learning/inferentia/

#reinvent

Tweet 75/104  Next tweet  Start

we’ll get there in a minute (I’m sure) but here’s more on @awscloud Trainium, the custom chip for #ml training

https://aws.amazon.com/machine-learning/trainium/

#reinvent

Tweet 76/104  Next tweet  Start

back to inference, @awscloud is working to improve inference work on general purpose CPUs as well. sometimes you just need to run the process where the workload is w/out the custom chips

#reinvent

Tweet 77/104  Next tweet  Start

…or at least _those_ custom chips

#reinvent

Tweet 78/104  Next tweet  Start

sticker stats for Trainium…

#reinvent

Tweet 79/104  Next tweet  Start

Peter says the name “Trainium” implies training #ml models. for most, it’s probably

#reinvent

Tweet 80/104  Next tweet  Start

“..as is normal in a math party” << f–k yeah, Peter

#reinvent

Tweet 81/104  Next tweet  Start

Peter diving into the math of #ml training runs. you need it to figure out how to scale out/up a workload to get the job done

#reinvent

Tweet 82/104  Next tweet  Start

GPT-3 is slightly larger than BERT-Large #nlp #ml models

#reinvent

Tweet 83/104  Next tweet  Start

these models are massive. they create their own computer science problems

#reinvent

Tweet 84/104  Next tweet  Start

more on how to tackle these challenges the @awscloud Well-Architected Framework, Machine Learning Lens

that’s available at https://docs.aws.amazon.com/wellarchitected/latest/machine-learning-lens/machine-learning-lens.html

#reinvent

Tweet 85/104  Next tweet  Start

more math party analogy/reference!

#reinvent

Tweet 86/104  Next tweet  Start

…lots of room left in that graph…

#reinvent

Tweet 87/104  Next tweet  Start

Peter talking about the biggest @awscloud EC2 instance types for #ml training

#reinvent

Tweet 88/104  Next tweet  Start

the network optimized version of the Trn1 instance type will have 1600 Gbps network bandwidth << fantastic!

#reinvent

Tweet 89/104  Next tweet  Start

sizing your training pool is a critical decision. not only will it impact how LONG you need to wait but also how MUCH you pay

read that Well-Architected Lens, please!

https://docs.aws.amazon.com/wellarchitected/latest/machine-learning-lens/machine-learning-lens.html

#reinvent

Tweet 90/104  Next tweet  Start

Tranium uses a technique called “Systolic Array Manipulation”

going to have to dig into this one later on: https://en.wikipedia.org/wiki/Systolic_array

#reinvent

Tweet 91/104  Next tweet  Start

Tranium provide 16 fully programmable inline data processors. this allows further tuning for each training workload

<< still doesn’t fix the name though

#reinvent

Tweet 92/104  Next tweet  Start

now we’re talking about how to round numbers. I ❤️ this stuff

super niche but fascinating

#reinvent

Tweet 93/104  Next tweet  Start

of course Trainium supports stochastic rounding directly in hardware. because, why not?

#reinvent

Tweet 94/104  Next tweet  Start

more on the @awscloud Neuron SDK at https://aws.amazon.com/machine-learning/neuron/

<< it abstracts a lot of what Peter just covered away, so you just get the benefits

#reinvent

Tweet 95/104  Next tweet  Start

Peter is going to wrap up with sustainability and the climate pledge

more on the pledge (now over 200 signatories) at https://sustainability.aboutamazon.com/about/the-climate-pledge

#reinvent

Tweet 96/104  Next tweet  Start

. @awscloud is always looking to improve efficiency in every aspect of our infrastructure << we’ve heard various achievements and efforts year over year from them. it’s nice to see

#reinvent

Tweet 97/104  Next tweet  Start

that’s a big commitment…and they are on a path to achieve it 5 years early (originally target was 2030)

#reinvent

Tweet 98/104  Next tweet  Start

#reinvent

Tweet 99/104  Next tweet  Start

storage isn’t the only problem. moving that power around is a big problem as well

#reinvent

Tweet 100/104  Next tweet  Start

here’s a great research piece from @voxdotcom

https://www.youtube.com/watch?v=s3ScJ_FwaZk&feature=youtu.be

#reinvent

Tweet 101/104  Next tweet  Start

. @mikegchambers called this the other night in our day one recap!

#reinvent

Tweet 102/104  Next tweet  Start

you can watch that recap on @mikegchambers excellent YouTube channel at https://www.youtube.com/watch?v=FiJfYTqM02M&feature=youtu.be

#reinvent

Tweet 103/104  Next tweet  Start

here are some other sessions at @awscloud #reinvent that focus on sustainability. be sure to check them out

#reinvent

Tweet 104/104  Next tweet  Start

..and that’s a wrap from Peter. not nearly as broad as previous years but still a great deep dive into things like @awscloud custom silicon

/ #reinvent