Adam Selipsky’s Re:Invent Keynote Left Us Wanting More (Live from Las Vegas, with the Finops Foundation)

Transcript

Speaker 1:

Okay, here we go.

Hilary:

Adam Selipsky in the house. Oh, my gosh. Starting out with Guns and Roses.

Speaker 3:

Hello everyone, and welcome to AWS re:Invent.

Hilary:

Live from re:Invent in Las Vegas. This is AWS Insiders, an original podcast by CloudFix about the services, patterns and future of cloud computing at AWS. CloudFix is a tool that finds and implements 100% safe, AWS recommended cost savings. That’s fixes, not just analytics. I’m Hillary Doyle, joined as always by Rahul Subramaniam, and we will be joined later in the show by special guests from the FinOps Foundation, J.R. Storment and Joe Daly. Hey Rahul.

Rahul:

Hey Hilary, and welcome to your very first live re:Invent.

Hilary:

Thank you.

Okay, Rahul, Adam Selipsky has just left the re:Invent stage. There is so much to discuss and I know it’s been a real emotional rollercoaster for you. We’ve been watching his keynote with friends and guests hanging out in the suite that Adele kindly left for us. As I mentioned, it has a sauna. It also has a dipping pool and a bathtub that I now live in. We’re in the Caesars Palace compound. And when in fake Rome, this is how Rahul Subramaniam hosts the watch party. But before we get into it, you watched Peter DeSantis’ keynote last night. I missed it because I accidentally took our team to a sex show. To be clear, this is not how the show is advertised, but would you please briefly give us your top takeaway from DeSantis?

Rahul:

Yeah, so I mean, watching Peter’s keynotes is always fascinating for me as an engineer. The two big things that came out of the keynote yesterday, the Graviton3E instances, the new chips that they’ve created, the new instance types that they’re now bringing in. We of course had the C7g, which was the Graviton3 instance. Now you have the new C7gn. Oh man. This is going to be twisting in my tongue for a while. That’s really cool. Great for network appliances. And then you’ve got the new HPC computing based instances that were announced yesterday. So yeah, a lot of very interesting engineering oriented announcements yesterday.

Hilary:

Great. Okay. Back to story time with Adam Selipsky. We know, we know that Adam appreciates the epic theme. This year was no exception. Global unrest and recession are obviously top of mind. And Adam explored the theme of exploration and how to survive and thrive whilst exploring. In other words, the economy may be terrible, the world is falling apart, but that does not mean you should stop spending innovation’s got to persist. One notable early quote from Adam, “If you’re looking to tighten your belt, the cloud is the place to do it.” There was a usual fire hose of announcements, serverless everything, better performance, easier data integrations, lower cost, heavy focus on security, ML, federated queries, data, supply chain. Did anything surprise you about this morning?

Rahul:

To be very honest, I thought Adam missed the opportunity to make a lot of very interesting announcements around how customers could actually leverage all of this stuff. It felt like a keynote that was steering itself to address arguments against the cloud, and he was just trying to tighten every argument that customers usually make about moving their workload to the cloud, but offered very little guidance on things that would made them definitively move to the cloud or move to AWS in particular. And it was a lot of storytelling that in places didn’t even seem to connect very well.

Hilary:

We were in Antarctica and I felt like that that was an unusual pivot, was not prepared for the tone shift. You’ve been very vocal about wanting to hear more about AWS taking hold of how customers should be using their products. Are you seeing that culture shift at all?

Rahul:

I think today they had a special focus on their supply chain tool, which is brand new. And they brought back AWS Connect where they’ve taken a vertical problem and created a solution very specific to that problem. And that’s how they are packaging up their opinions because now they’re literally telling customers how they should manage their supply chain. And they’ve already been telling their customers how they should be managing their contact centers, but without actually telling them how they should be managing their contact centers. So I think this is the middle ground that kind of satisfies AWS’s culture as well as helping customers navigate this super massive space of AWS services.

Hilary:

Top product, top announcement for you coming out of this keynote.

Rahul:

I wish they had done more to address customers’ concerns about the cloud being more expensive in their view, but I think the push to serverless was definitely the underscored theme that they were continuing from last year. They obviously had a lot of the Aurora Redshift, all of them move to serverless last year. And now with the last piece, which is open search also offered at serverless, their entire analytics platform moves there. And so it really helps customers scale up and down elastically.

I think what Adam was trying to communicate to the customer base was the only way to be cost effective is to try and adopt elasticity in your… [inaudible 00:05:34], predicting exactly what you will need and provisioning it for it in advance. The more agile you are in terms of consuming all the resources, scale up when you want, scale down when you want. He said that about three or four times. But yeah, I think that is the message to the customers. Leverage elasticity, build your stuff to be elastic in nature and the more serverless you get, elasticity becomes really easy to leverage because AWS does all the hard work for you.

Hilary:

Got it. Rahul has spoken, let’s open up the floor. AWS has helped popularize the concept of cost optimization, but more and more cost optimization can’t be divorced from FinOps. So we have brought in the kingpins of FinOps, the godfathers of the FinOps Foundation.

J. R. Storment is the executive director at the FinOps Foundation. Joe Daly is the foundation’s director of community. The FinOps Foundation is a financial management, discipline, and cultural practice for the cloud. Basically, the Foundation helps to pull engineering, tech, finance and business together to help maximize efficiencies across speed and savings and quality. I mean, who does not want that? Hold onto your hats people. Welcome to the podcast JR. And Joe.

JR:

Thank you Hillary. She mentioned holding onto our hats because we’re literally up here on a roof and the wind is blowing.

Hilary:

Yesterday, I got hit in the head by that sign over there with our faces on it.

JR:

It’s a nice view though.

Hilary:

Yeah, it is a nice view. It’s worth it for the wind. You guys focus hard on the economics of cloud computing. There were fewer direct cost optimization announcements this morning, which was a surprise to us. More of an assumption I guess, that if you’re with AWS then cost optimization is just built into the package. You’ve already got it. Should business believe this?

JR:

Well, I think the beauty of cloud is you only pay for what you use, but the challenge of cloud is you pay for everything you use, whether or not you’re actually getting value out of it. So I think to Rahul’s earlier point, there was a lot of storytelling, which I saw really as them opening up new markets into which they could sell software. There was less specific announcements as we’ve seen. And this is my ninth re:Invent. I miss the days of them driving semi-trucks on stage when they were announcing Snowball. There used to be these big, we’re doing this amazing new thing and this one felt like a lot of integrations and fine tuning in terms of product announcements. But really now we can do omics and bionomics and protein omics, in various areas. So no.

Hilary:

Talk a little more about that. Okay. So this is interesting. I mean this is my first re:Invent and I think for a lot of our listeners they’re, they’re learning about re:Invent in real time. Why do you think this feels like a toned down experience for you this year, Joe?

Joe:

To use Rahul, I think you said it was a missed opportunity and I think it was a missed opportunity honestly. They have great potential and I just think in this time, in this place where there’s a lot of economic uncertainties that just hammering down on here is how you can increase the economic value by leveraging our services and just hammering that point home on every announcement. I just assumed they were going to do that. But I mean did get a really great lesson on Antarctic exploration.

Hilary:

Yeah, I know. [inaudible 00:09:01], what a guy. I mean he’s who you want on a desert island.

Joe:

I mean I want him on my FinOps team.

Rahul:

[inaudible 00:09:06]

JR:

It felt like a thinly veiled jab at Microsoft or Google or Oracle, which was basically the story of the winner takes all. And the story for those who didn’t see it was that there were two teams racing to the South Pole and the team that won not only got remembered for it, the other team basically died on their return and are not remembered. And it was sort of-

Hilary:

Except they’re both named at the South Pole, which I felt like-

JR:

Consolation prize. But I think the thing that stood out, and it was the consistent from all the keynotes… There’s always the announcements… You were mentioning all the new instance types, right? I’m not even remembering all the new acronyms and the common theme that 4x more performance, more throughput, which ultimately speaks directly to the cost optimization piece, move to these. And I think fewer blockers now to move to them. Some of the announcements last night, they were talking about more IO and throughput and some of the size instances that historically didn’t have it, which gives you more abilities to optimize as you go.

Rahul:

Yeah, I think there was definitely an opportunity to separate out, here is how you optimize your existing stuff and here’s how you innovate. And that would’ve been a great way to separate out that keynote. You can say, here’s how you optimize your current spend, take away all these savings and then go invest it in all this new innovation that’s going to be really exciting for your business, especially in the current economic landscape. And I think that was the big missed option to separate that out, help give customers greater clarity about what they should do.

Joe:

I always find these announcements the most… What they do most often is increase the cost of my tech debt because there’s all these great services, all these serverless things. And when I was a practitioner and I had just hundreds and hundreds of different types of applications, I was helping do the FinOps management with those teams, there were teams that could take advantage of the services as soon as they were released. And that was really exciting and the great stories and is great value. But the vast majority of the application teams I was supporting are doing IAS. It’s hosting. So I mean think the most exciting announcement I heard from that perspective was IFN instances with the high throughput. I’m like, oh, I could host databases a little bit cheaper.

JR:

The big thing that stood out that I was really happy to see them talking about was all the data analytics and making the accessibility of those things better. Because underlying the entire FinOps discipline is visibility and collaboration, accessibility of data. In order to drive more accountability for cloud spend, in order to drive better investments, you can innovate where needed, save where needed in all those areas. And so the announcements around… Well first of all, the serverless stuff for Redshift and those things were great in terms of the leverage you can pull to turn things on and off, but that you could start to do no ETL Redshift with spark queries and all these different things plugged in. To me it was like, okay, great. Well there’s a better way to start accessing a lot of the data and then push that stuff into some small announcements around QuickSight.

Oddly, the ability to print QuickSight dashboards was like, yes. Because part of the issue with FinOps data is that you need to get it out of the cloud and out of… If you’re using FinOPs platform and get into the path of the engineer, or in the case of the printing the QuickSight dashboard, get into the path of the executive. And that’s been a big challenge is getting that data out. So a lot of what they’re doing is… We always joke that every time AWS makes announcements about more features and more granularity and per second billing or millisecond, it makes FinOps more and more relevant because the challenge gets harder, the data gets bigger, the confusion gets deeper. So while they give us more levers to pull, they’re also adding to the complexity the practitioner has to deal with.

Hilary:

Do you think that that’s the hardest piece for FinOps applicability or FinOps pickup, right now?

Joe:

The more services that there are to… Well my goodness, what they said, there are 600 instance types.

Hilary:

That made me sweat.

Rahul:

I was like, last year was 450 and this year is like 600 plus.

Joe:

In reality, most companies probably need eight. And so you have 600 choices. Your team’s going to need to choose which ones are we going to enable for our organizations to make sure that they’re secure, patchable, and that they meet all of the compliance of that specific company. So I mean, it’s fantastic to have a wealth of options, but it can also lead to the inability to make a decision and what should we do? What’s the best thing to do? And it’s really hard. So you need that data to be able to make choices. And that’s the beauty… My favorite thing about the cloud is the amount of billing data you can get and you can really start to iterate and learn from your POCs and you test, you’re like, oh, this is good.

JR:

And that’s the nirvana state of FinOps, which is the data driven decision making by the teams at the edge. But then ultimately, I mean to the question of how to intersect with FinOps world, I mean it’s funny, we pulled some numbers recently about companies participating in the FinOps Foundation, you mentioned the 86%, we just ran on the Fortune 50. There’s 86% of the Fortune 50 in the FinOps foundation right now. And so we’re seeing a real mirroring obviously because these companies are dealing with the complexity at ABS, but also dealing with the complexity of Google and Microsoft and all the other services out there. Because the enterprises have this huge multi-cloud billing data issue in our world, which is how do you first of all even understand it. And then how do you start to marry together, normalize it, and then make those decisions that Joe’s talking about.

Rahul:

You got to keep at it and you keep getting better at it. I mean, if the external variables stopped evolving, which is if AWS stopped doing new services, if our use of AWS just stopped progressing in some manner, I would say yeah, we’d probably be static and we could probably achieve a certain state of maturity and kind of stop at that. But everything is changing. The economic scenario is changing. AWS services are changing. The way we use AWS services is constantly changing because we are evolving. And like I said, I mean there is so many announcements being made on basic bare bone infrastructure like instances. But then in my world, all the values actually in moving up to the higher order services where things can get orders of magnitude cheaper and more efficient and effective, starting to use serverless… You have to move up the stacks and people are going to be in different places across their portfolio, across their applications, across people [inaudible 00:15:32].

JR:

Yeah. Ultimately. That was great to see so much of the focus on their moving away from talking about infrastructure as much and everything is now serverless underpinned, which-

Rahul:

Correct. Actually we missed out one of the big announcements, which is a no ETL future. They took something that is manual, mundane, and painful for the practitioners and putting all the data together is one of the hardest things, which is why you’re not able to leverage most of it and they’re doing… Having that as a vision, I thought that should have been really, really big and should have been the whole team-

JR:

That data service catalog was-

Rahul:

And it got lost in the story.

JR:

Keep me honest on the no ETL thing though. The data needs to be in Amazon first, right? They were talking about no ETL between Redshift and Aurora, right? And it felt like… And they got into running Spark on there and then the Q thing in Redshift… Or sorry, in QuickSight, which lets you query the data. It felt like they were in that part really moving away from speaking to the technologists and starting to talk to the business people, which is, hey, you want to ask a question of your data about the top 10 biggest spenders or biggest things, you can do that really quickly and easily. And that was kind of… In that way, I think the story worked really well, which is we’ve got your data, we’re making it easy, now we’re going to give the business insights you need.

Rahul:

Correct

Hilary:

JR. And Joe, thank you so much for being with us.

JR:

Thank you so much, Hilary.

Rahul:

Thank you so much for coming. This is so much fun.

Hilary:

Okay, Rahul, for the next three days of this epic re:Invention, we are taking audience questions for you.

Rahul:

re:Invent, not re:Invention.

Hilary:

No, I was saying it’s an epic re:Invention.

Rahul:

Sorry.

Hilary:

Yeah, like today’s epic exploration. It was word craft, Rahul. And we’re going to start with this question from Lauren. Your guest yesterday talked about there being a need for different chips for different instances. Basically the idea that there might not be a one size fits all solution in the future. What do you think of that? Lauren is referring to what we heard yesterday from Brad Smith at AMD. Lovely guest. Rahul, no one size fits all in the future. Does that sound right to you?

Rahul:

Yeah, I think the big movement that AWS has kind of pushed for is to have the right technology for the right job, instead of having good enough. I think that’s something that Adam talked about today.

Hilary:

Yeah. Good enough is not good enough,

Rahul:

Good enough is not good enough. And Andy Jassy has actually been talking about it for about five years now.

Hilary:

You’re really missing Andy Jassy. Are you okay? Should we take a moment for you to-

Rahul:

Just as a technologist? Yeah. I loved Andy’s keynotes. It felt like Andy Jassy was directly addressing technologists and he would… The entire keynote was kind of geared towards that. Adam’s feels more marketing oriented, but the point that AWS has always been trying to push is when you get specialized to build a certain kind of tool to address a certain kind of problem, you can get cheaper, you can get faster, you can get more scalable and more elastic at the same time. But if you trying to build a jack of all trades kind of tool that try to do a little bit of something, little bit of something else, you never do anything really, really well. You do everything at a very average level.

And I think that is something that AWS has enabled. Now historically, when you’re OnPrem, you want to just have one thing to manage. But as we move to the cloud, that whole constraint kind of goes away because you’ve got these manage services that you can pull in. So you can have an OLTP database for your transactions. You can have an OLAP… I mean the entire analytics platform that they talked about today to do just analytics. You can separate out those concerns and do an amazing job with both of them independently and not have to be constrained by one tool, say Oracle, that tries to do everything for you and doesn’t succeed in doing anything really well.

Hilary:

95 people in the lobby just took a shot. Before we go, would you give us a look ahead for tomorrow, one thing you’re expecting to hear or would like to?

Rahul:

Okay. So the buzz is that there’s going to be tons of new announcements around ML and AI stuff during Swami’s keynote. My guess is there’s going to be tons of conversations about managing your entire ETL pipeline. There’s going to be a doubling down on the focus on being ETL free as a future. And so the more machine learning and analytics and AI that comes into managing all your data, understanding what it means automatically and being able to merge them all together and integrate them together, that’s going to be a big theme in the keynote tomorrow. And then of course, all the new [inaudible 00:20:06] that just allows you to automagically create all these insights from your data. That’s going to be a big part of it. So I know all the machine learning and AI teams that I’ve spoken to, again, they remain very tight lipped, but you can make out that every team has something that they’re anticipating in the announcements that are going to come tomorrow. So I’m really looking forward to Swami’s keynote tomorrow.

Hilary:

Okay. You heard it here first. That’s it for now. Much more to come. We will be reacting to Swami Sivasubramanian’s talk in tomorrow’s special episode. Episode three, can you believe it?

Rahul:

And in the meantime, please send us your reactions to podcast@cloudfix.com.

Hilary:

Rate, review. Five stars please only. And yeah, see you tomorrow.

Rahul:

See you tomorrow. Looking forward to it.

Hilary:

Me too.