Mission Talks: Andy Jassy re:Invent Keynote 2020 Insights [Video]

In this video, Mission CTO Jonathan LaCour, Consultant Tim Banks and VP of Consulting Services Jaret Chiles discuss the highlights of Andy Jassy’s 2020 re:Invent keynote.

Transcript

Jonathan LaCour:
Welcome everybody. Thanks for joining us today. My name isJonathan LaCour, I'm the Chief Technology Officer at Mission. I'm joined bysome great, smart folks from Mission, as well. I'll let them introducethemselves real quick.

Tim Banks:
I'm Tim Banks, Consultant here at Mission.

Jaret Chiles:
And Jaret Chiles, Vice President of Consulting Services.

Jonathan LaCour:
Awesome. So we're here today to kind of have a little bitof a, just a chat, casual conversation, reacting to the exciting Andy Jassykeynote this morning from re:Invent. And there are so many topics that we couldcover. This is not going to be a summary of everything that happened. We're notgoing to tell you about a list of all the announcements that were made, becausethere were a lot.

Jonathan LaCour:
Really, we just wanted to do something a little bit morefun, and just chat with each other, and talk about what kind of stood out tous. And for a little bit more context, the reason this is interesting to us is,well, we work for Mission. Mission is a consulting and managed service partnerof Amazon. We're actually a premier consulting partner, and managed serviceprovider. So we are 100% focused on AWS and therefore, we were watching prettyclosely this morning.

Jonathan LaCour:
And I'd like to actually start with just the theme of thewhole thing. Andy Jassy really leaned into this concept of what does it mean toreinvent? And I thought that was really, really interesting.

Tim Banks:
Yeah. I thought it was interesting, as we talk about reinvention,and the first company that came up was JP Morgan Chase. And you think of an oldfinancial institution that's been around since folding money's been around,almost, that is using modern technology to reinvent how they do their ownbusiness, and how they service their customers. What does the financialinstitution, what does their company look like two years, 10 years, 20 yearsfrom now? And they're really looking at changing who they fundamentally are tomeet those needs.

Jaret Chiles:
Yeah, the thing I would add there, too, is the emphasis onbeing proactive about reinventing. He made some very specific points about it.If you wait until you need to reinvent to reinvent, you're way behind already.

Jonathan LaCour:
You've already lost.

Tim Banks:
Yeah. And I think, too, you're talking about how they did reinventionideas. They mentioned a company called Boom, who's doing transatlantic... Well,not transatlantic, but just supersonic flight. Supersonic flight has beenaround since, I guess, since the 60's. Back in the day as the Concord. And ithad gone away, because it was no longer sustainable for doing. But Boom istaking that idea, and they're bringing it into the 21st century, and saying,"All right, well, we can find a way to do this that's cheaper, that's moresustainable. And so we're going to take this old idea, and we're going tomodernize it, we're going to bring it to the masses."

Jonathan LaCour:
Yeah, but do I still get like a little flute of champagnein the Concord kind of way? You know? I want it to be like it's somethingfancy.

Tim Banks:
No, I think they just give you those little mini Diet Cokecans now.

Jonathan LaCour:
Yeah, it's a shame.

Jaret Chiles:
If you can get me there in four hours, I'll take it.

Jonathan LaCour:
Yeah. I think AWS does like to talk a lot aboutdisruption, reinvention. But frankly, they kind of live up to it. And I alwaysfind that really interesting to see how they're willing to do some things thatyou think would cannibalize themselves. And they're just okay with it, becausethey know it's the future. It's like always looking forward, always lookingahead.

Jonathan LaCour:
I like to think that Mission does the same thing. We'realways thinking about making changes and adjusting and such. And it's reallygood to have a partner like AWS that kind of pushes you to do that. Goodreminder.

Jonathan LaCour:
And I think, go ahead Tim.

Tim Banks:
So I think it was interesting, too, we talk about reinventionkind of disrupting. And one of the things I thought was AWS is definitely goingafter some of the larger competitors and disrupting them. They talked about...They had taken several shots at Microsoft with some of their roll outs. One inparticular was Battle Fish.

Jonathan LaCour:
Yep.

Tim Banks:
So they specifically have this sort of open sourced thingto help people get off of SQL Server, and onto PostgreSQL. That is really avery... That's a very, very pinpoint shot.

Jonathan LaCour:
Yeah.

Tim Banks:
That's a sniper shot right there, you know?

Jaret Chiles:
It is not a small undertaking to re-platform yourdatabases. I mean, when you can remove that friction and make that seamless andreduce the licensing costs and everything.

Jonathan LaCour:
And we talk to customers all the time where they're like,"I'm trying to cost optimize, and I have this huge cost, and it'slicensing." And it's like when you get a call from the roofer and you needto get a new roof. Nobody wants to get a new roof, because what value am I gettingout of this? Same thing as licensing. I feel like I'm not getting the value.And when there's great open source relational databases that could work prettywell for my use case potentially.

Tim Banks:
And what I think is interesting is they didn't say there'sno lift.

Jonathan LaCour:
Right.

Tim Banks:
But they definitely decrease that lift. Whether you're onOn-Prem or whether you're using RDS and SQL. They're trying to get you out ofpaying licenses, and them out of paying licenses to Microsoft, to access yourdata.

Tim Banks:
What I think is interesting about that is even thoughthere's going to be a lift, it's a lot easier. If you don't have the cycles todo that, that's something that you can farm out. And something Mission hashelped with many times, of getting people to do database transformation ontoRDS or onto Aurora. We've got plenty of expertise and have done it all daylong. So I'm excited to see that, because it's going to help people save money,it's going to help people do things a better way, it's going to help people bemore elastic, and be scalable without having a lot of heartache. And alsowithout having to be subject to the whims of licensing.

Jonathan LaCour:
Yeah.

Jaret Chiles:
That's probably a good segue into the Aurora announcementsin general. What do you guys think about version two?

Jonathan LaCour:
Well, Aurora Serverless version two is very exciting,because Aurora Serverless version one was not very exciting.

Tim Banks:
I think Aurora Serverless version one was kind of like anexperiment that just went to production. And it had so much potential. Hugepotential, but it just never delivered on it. So I'm excited to see that withAurora Serverless v2. When we hear the claims that it would scale up fromhundreds of hundreds of thousands of transactions in fractions of a second,that's a big claim.

Jonathan LaCour:
Huge claim.

Tim Banks:
That is a big claim. And if it works, I'm here for it. ButI definitely want to see it. Because if you think about it, how long does ittake to spin up an RDS instance?

Jonathan LaCour:
Oh yeah, exactly.

Tim Banks:
Minutes, right?

Jonathan LaCour:
Minutes.

Tim Banks:
And sometimes minutes is way too long.

Jonathan LaCour:
Oh yeah.

Tim Banks:
If you're talking I can scale up transactions across AZ's,and I don't have to worry about replication, because it's just compute. Andyou're talking even faster than even some of the land [inaudible] problems wehad in the past, that's really compelling. That's really compelling.

Jonathan LaCour:
I'm an old school database nerd. I love relationaldatabases, and I've lived through the whole document databases, and all thiskey value stores. And it's all great, I love those, too. But there is aspecial, wonderful thing about relational databases and SQL, which has beenaround for ages. It is truly fantastic modeling tool. It's incredible, powerfulthing. And it's always been sort of like, well, we have RDS, and it's basicallyjust like the old thing, but just wrapped in a couple EC-2 instances auto-scalingand managed service around it, and it didn't really feel like it was thisgenerational leap. And to me, that's what Aurora Serverless v2 couldpotentially be, hopefully.

Tim Banks:
It's always been the long pole in the tent. Rapid auto-scalingon your database back in, and this is the answer, right? We'll see.

Jonathan LaCour:
Yeah. And I guess it does not, is not lost on me that theyalso had a lot of storage-related announcements which we talked about, as well.And that's what kind of this scaling problem really comes down to in manyrespects, is it's a storage-scaling issue. So I'm guessing these teams areprobably talking to each other.

Tim Banks:
Yeah, it's interesting though, because Amazon announcementthe GP3 storage for block storage. Which decouples IOP's from storage space,and that was always a problem. It's like hey, I need more storage, I need moreIOP's. I need 10 times more IOP's. Well, now you have 10 times more storage. Ididn't need 10 times more storage, I just needed the IOP's, and now you gotthem both.

Tim Banks:
But it goes back to when we were talking about withAurora, now we can have almost as much storage as we need without having to addcompute to it that you weren't going to use. So literally, it's the opposite,but still so necessary. Because now people can do some optimization. How manycustomers do we have that have so many volumes just sitting out there almostempty, but because they needed the IOP's, they have to have them? And there's abig potential there to optimize, to get rid of all those, to get a little moreefficient.

Tim Banks:
And that may be a big lift, and that's fine. Somebody'sgoing to pay dividends in the long run, but if you're out there, and you'relooking at your volumes and your console, and you're seeing them all sittingthere half unused, but you need the IOP's, and you're figuring out well, howcan I get these consolidated down into GP3 instances, let us know. We can helpyou with that.

Jaret Chiles:
There has been a lot of cost optimization exercises offthe back end of this GP3 announcement, that's for sure.

Jonathan LaCour:
Yep. All of our Cloud Analysts who do cost optimizationservices are going to be spending the next two weeks looking at everybody'svolumes and saying, "How can we save?" It's another lever, though,right? That's the whole thing that AWS tends to do is they tend to take asolution, and they figure out how many turntables they can give you as thecustomer to give you the power to do literally whatever you want with it. Whichis awesome, but the downside of that is it's a lot of complexity. Whichselfishly, that's why we're in business. Partially, at least, because we canhelp with that. But it is really interesting.

Jaret Chiles:
A more more options is a lot more to solution with. Imean, it's a big toolkit, and you need people that know how to navigate itquickly.

Jonathan LaCour:
Exactly.

Tim Banks:
I will say, speaking of options. The thing that I'mexcited about is some of the silicon they're coming out with EC2.

Jonathan LaCour:
Yeah.

Tim Banks:
Like AWS has been using the Graviton, and that's beenawesome. Now they're talking about using the Tranium silicon. This is twodifferent now types of custom silicon that AWS has themselves.

Jonathan LaCour:
It's three now.

Tim Banks:
Three now.

Jonathan LaCour:
First one they did was Inferentia.

Tim Banks:
Infrentia, yeah.

Jonathan LaCour:
Yeah.

Tim Banks:
And so you've got that. They're offering MacOS instancesnow, and I don't think it's any accident that they now have ARM-basedinstances, and now they have MacOS that's now compatible with ARM, as well. Sothese are some pretty awesome kinds of things that we have options for, and wealready know that it's a 20% savings off the top for Graviton processors rightnow. So utilizing some of the Tranium, that could be really, really lucrativefor some folks.

Jonathan LaCour:
Well, it's interesting to me, because the first kind ofround of this was sort of signaling. But now we see where it could go, and youbrought up MacOS running on AWS, which is super interesting, when they didthat. But you can actually do a comparison to what Apple's doing in theirsilicon, with their consumer kind of PC's, and their phones, and their tabletsand everything.

Jonathan LaCour:
And AWS has more workloads running on it than any otherCloud provider in the world. They're customer obsessed. This is their big,cultural value that permeates AWS. If they're able to develop custom siliconthat matches to what they're seeing with workloads in AWS, watch out, becauseeverything that they've done up until this point, until Graviton and some ofthe other things, has been based on commodity compute from Intel and AMD.

Jonathan LaCour:
And this is a totally different ballgame now. Customsilicon for all sorts of different workloads, and right now it's for machinelearning, and that's the logical place for them to start. But yeah, it'sinteresting.

Tim Banks:
No, yeah, it really is interesting, and I think some ofthe things that you talk about with AWS really talking about disruption astheir whole theme and reinvention as a whole theme, they're literally reinventinghow they do silicon, like what their instances look like. And if you talk aboutthe work that they're doing right now, and the work that the companies can doright now to utilize some of the new technologies coming out, we're talkingabout something that's going to apy dividends for years. Whether it's costoptimizations, whether it's performance, whether it's some of the AI technologythat they've been talking about, some of the analytics tools they've beentalking about.

Tim Banks:
A lot of the things they're doing now is a spot where AWSis reinventing itself so that customers can reinvent themselves. So they can bein a great position to really take hold and grow on some of these platformsthat they're coming out with now. And who knows what that's going to look liketwo, three, five years from now? I'm super excited to see it.

Jaret Chiles:
Well you've already started seeing some of that. They'rebaking some of these services in, these analytic services into their connectportfolio, and they're manufacturing hardware, and all these things. I mean,it's just a fly wheel of agility and momentum that they're gaining with theseplatforms that they're able to use to service their other features, as well,and it's just mind boggling to think where this is all heading.

Jonathan LaCour:
Yeah. And we're at the beginning of this thing, too, whichis really terrifying in some respects.

Jaret Chiles:
I know. Just scratching the surface, really.

Jonathan LaCour:
All right. I do have a topic that is near and dear to myhear that I want to talk about, because I still consider myself a developer,even though I don't get to write code nearly as much as I'd like to. And I lovewriting serverless applications, and I like, also, building out containerizedapplications. And for the longest time, it's been a pain, because you have oneset of tools that you use over here, and one set of tools over here. And twodifferent workflows, how do you do CICD? And developer experience has neverbeen one of AWS's number one strengths. They build out these incrediblebuilding blocks and atoms, and then they kind of just leave it to the communityto figure out some of the other bits and pieces.

Jonathan LaCour:
But they did announce the whole you can build out LAMBDAfunctions from container images now. Which is super cool.

Tim Banks:
As a dev ops guy, and somebody with an operationsbackground, the idea of only having to maintain one tool set or deployingLAMBDA functions like I would any other doctored container is a huge gamechanger.

Jaret Chiles:
Oh yeah.

Tim Banks:
And complicating what you do. And this is just me doingconjecture. What I would love to see is this turn into something where you candeploy LAMBDA functions in something like Kubernetes. Where I just have a podto find that's nothing but LAMBDA functions, and I can schedule them where Ineed them, auto scale them or whatever. That would be pretty awesome. I don'tknow. I'm not saying that's going to happen, but that's what I'd like to see.

Jaret Chiles:
Is that a prediction? We'll see.

Jonathan LaCour:
You just broke. It is really interesting though. I'm a bigfan of the serverless framework. I actually think it's really great and reallyuseful. But I'd be interested to hear their take on this. Because look, it'spossible that it's sort of a second-class citizen. Like, that's one of thethings that I think remains to be seen. I haven't dug in enough yet. Are therelimitations if you deploy a LAMBDA from a container image, or is it just aLAMBDA? I don't know. I'm very intrigued to see what the community sort of saysabout this. This is really interesting.

Jaret Chiles:
It's one of those things. It's just more tools in the toolchest. It's a good option to have. But it does, it increases complexity in itsown way. In one way you're reducing complexity by maybe simplifying the numberof tool kits that you have, but now it's just more options to choose from. Youneed partners and folks that have been through it that can help you wadethrough. It seems like everything, there's always pros and cons. And so stillto be determined what we're going to find from some exploitation of thesetools.

Jonathan LaCour:
Of sure. And the good news is I know a guy. So if anybodyneeds some help, just let me know. I know a guy who can help you with all yourAWS needs. All right, gentlemen. That was a really fun conversation. Wedefinitely didn't touch on everything, but we touched on some of the reallyinteresting and fun things. And looking forward to doing more of these over thenext couple of weeks, now that AWS has decided that we need to reinvent our livesfor three weeks instead of just for a couple days.

Jaret Chiles:
Hey, I'm in. It'll be fun.

Jonathan LaCour:
Yep.

Tim Banks:
It'll be great. Good times.

Jonathan LaCour:
All right. Good stuff. Thanks, guys.

Jaret Chiles:
All right. Looking forward to more. Thanks.