Liam Randall, Cosmonic & Colin Murphy, Adobe | Open Source Summit NA 2022

This is Textron TV. Hey everyone. Welcome.

This is well. So day three of our coverage for the open source Summit, but we started a day early. So it's actually day two officially of Linux foundations open source Summit here in in Austin and I'm really happy to be joined by two friends of mine.

Well one's a new friend ones in old friend. It's not that he's old, but I know a long time let me first of all introduce you to my friend Liam Randall from cosmonic Liam, welcome. That's great to have you great to be back on text strong.

I love coming on it's always a great conversation. So excited about today. Absolutely.

What is the first time we're doing an in-person. That's right. We've done a few virtual the last few years, but it's great to be back together be back.

Yeah. Just real quickly before we jump into things. We are in person.

We were in person at RSA. And you know, we had a lot of folks unfortunately can check covid. I will tell you that I think overall the Linux foundation's done a really nice job here.

I mean, we're not wearing masks right now, but I have masks in my pocket and I'm sure they're both of you do when you're walking here. You're wearing masks. They're checking temperatures.

I mean look nothing's perfect. But I'm I was yeah, I I think the Linux Foundation truly Embraces their mission of being Community First, you know, it's an open source company it's built on a together and I think that I know that they work very closely with major stakeholders in their ecosystem to craft policies that put the safety and health of all the attendees first. Yep, so so kudos to them for that.

So we've got Liam Carl. I I'm blanking you Murphy Colin Murphy and Colin is with Adobe. Actually, why don't you introduce yourself time?

Go ahead Colin Murphy. I'm a senior software engineer at Adobe on the CC web team. So some upcoming creative products, but before that I was in document Cloud for a number of years.

Excellent. And of course CC web is the Creative Cloud right a web team. Yeah.

I'm a customer. But anyway. The reason we have two folks here is I wanted to discuss something.

That's you know, we've seen it kind of percolating here as open source Summit this year. Look, we are 15 years into the web. shift and lift Some of its Cloud native whatever.

But we are up, you know, just when we start getting our heads around that we're starting to see the The Edge right the Horizon that's kind of the new frontier and it promises amazing things. I had a discussion with Liam yesterday. We recently did an event Philadelphia of all places with a gentleman from Horizon who runs their 5G commercial practice and he was all about the edge.

Yeah, and it really opened my eyes to What a game changer. This is yeah, I mean the industry is really been driven forward by these enduring themes and Technology of the last 20 years. And if we think of them as layers of a cake, you know, the first was virtualization and that started with VMware and paved the way for the first epic of cloud computing but especially with containerization.

I love your term. We really were riding this great lifting shift into the cloud, you know, closing down data centers moving together virtualized assets, but the next epic of computing I think is clearly defined by this great opportunity to bring our Computing out towards the data towards the edge and the users and there's a huge array of factors that are lining up in order to drive that story regulation limited and deliberate autonomy a privacy performance the edges where the data is so we have all of these things strategically looking ahead and I think it's going to be the theme that comes the Dominate the next 10 years of computing. Absolutely.

So I'm a little older than you even before virtual. Should I remember when we first got the the TCP stack? That's you know what?

I mean? What I'm Revelation. That was my first project as a network engineer was migrating net buoy and ipx SPX the tcpi.

Yeah. Yeah. So, you know, I you know, if you haven't, you know routed more than one protocol in your life, you know, I think you absolutely good those it was hard but it wasn't that hard.

Yeah. When as I look back on anyway, we're gonna trade config that system. No, no, no your email 386.

Yeah. That's it. Come home now.

Yeah. I remember those that I had Venus. Anyway, let's not even go down that fast because we'll be here all day.

But another important thing I just want to make sure our audience recognizes is when we talked about data on the edge and users those users and that data is not necessarily generated from people. There, you know the io I I think one of the big things about the edges that's where these iot devices are going to live and the quote Carl Sagan, right? There's billions and billions of them.

Absolutely and that's gonna generate Mind boggling amounts of data. It does generate my boggling amounts of data. And where are we gonna process that store it analyze it work with it or yeah, exactly.

So that's that's the big thing and I'm actually gonna talk in about a half an hour about this is one of the main use cases that would be that's that we're looking forward to it. Adobe is is around collaborative editing and machine learning at the edge and you don't need to send everything back to the data center because that's expensive right? That's that's one of you know, we if you have if you're at scale everything, you know, every megabyte coming into your data center close your money, it costs money and and the compute and so if we can take advantage of edge where where network is cheap right and and low latency and then we can kind of send what we only really what we need back to the data center is the idea.

So I I think that is. One of the kind of trigger questions we need to be able to figure out which is what stays at the edge what goes back to the to the cloud and I'll give you something else. I've spoken in some companies who do sort of like data analysis on the Fly for for data.

There may be a thing where maybe you don't do we need to keep every bite of data whether it's on the edge in the cloud or on someone's machine, right? Are we are we kind of wearing belts and suspenders by keeping the amounts of data we have well, I I think it looks like you know a set of you know bandpass filters or you know sieves that you're passing data through, you know, and you think about like your nest doorbell, for example, you know, you're not streaming 100% of the video bandwidth back, you know on that device. There's clearly some lightweight machine learning models that are doing initial detection of for example, maybe a face and then when you identify those trigger, you know pieces of data that you filtered out those can get passed up to more advanced machine learning in the cloud to build it on Collins point.

I think it's a looks like a pyramid we are you know, we have an incredible array of data here, but we can only choose to pass and process so much across as we move up the stack and up to cost that too as we as we think about that analogy probably holds pretty well. Absolutely. so Colin spoke a little bit about the adobe's interest in this what cosmonic well cosmonic was founded on on the trying to figure out a way to build software easier on this enduring theme and a few years ago as a working at a large fi I was a a VP of innovation at Capital One and we looked across our technical portfolio and we saw an opportunity to really address how we build distributed software.

So we open sourced and started to build a real Community around what's now called Walzem Cloud on Walzem cloud is in the cloud native Computing Foundation. It's an incredibly fast growing open source project who's users include Adobe BMW, we've got huge contributions from Capital One, which Telecom and a huge array of companies that transcend industry and as a framework what it does is it takes this distributed Edge and it gives you the ability to write software that can seamlessly move across it. Transcends devices operating systems browsers CPUs, but it gives the the developer the opportunity to just work on their hard problem that matters their business logic BMW for example is using it to run machine learning models in their cars and what they love about it is it gives them the ability to pull those same models back into their kubernetes or their clouds of awesome Cloud operates compatibly with all of those things Nomad kubernetes Etc.

So cosmonic is built on top of Wasim cloud and is providing a platform as a service for that community of people that see this opportunity and are working to address it in their own organizations God. Just for you know, sometimes we talk and people are following this but this there's no Links at the bottom of our live coverage wasn't Cloud. Where can they get more?

com is has links to our community calendar. Today's actually Blossom Cloud Wednesday. So we have our community call it one Eastern today.

We list all of our conferences there. com where you can get Hands-On without installing any of this. We also maintain a really active developer community at rosem Cloud dot Dev and it's a part of the cncf.

There's transparent Community governance here. So you can see who are contributors are you can see who's involved where contributions are coming from and really understand the community and it's really a community that transcends our core team. There's less than 10 of us at cosmonic today.

However, there's over 120 contributors to the core of Lawson Cloud. So it's an idea that definitely transcends. And what we're doing?

I got to get a little Elementary with them for one second again, we're on video. Can you spell watch them for them shirts? Just w a s m it's it's a short for webassembly is abbreviated as wasm and I think webassembly is maybe one of the next things we could probably talk about because well that that's Segway just kind of thrown out here.

Yeah, but and we're gonna dive into it, but I've got to give Tom chance. Go ahead. Oh just started starting off on webassembly.

No, no, you were gonna say something before I asked me to spell awesome. Oh, no, I was just gonna say, you know, I don't even remember that happened. So yeah, I do that to people.

No, it's fine me. I was all geared up for webassembly. Oh you jump in the webassembly.

Well, I just want to again I don't want to put you on a spot with Adobe. But Adobe is a user was Some Cloud. Well, we're not in production.

We've we've we've definitely given some evaluations and we and we've had a lot of meetings with I mean we've had you know, we've been really productive. I was just kind of kind of saying the nice thing is that we have we have the this webassembly is part of the W3 right and then but then we have this this kind of we if we once we talk about that, we have to kind of talk about Wazee and the bytecode alliance and was in time and and that's where Liam's stuff is really that's what Liam stuff's really about and it's it's a really fantastic community. And so it's it's funny.

You have to kind of peel back this onion to get to sounds like yeah, we believe a couple more layers, but at the heart of it, is this webassembly, that's right. Yeah, look the industry has tried to address this right once run everywhere problem before and we've all heard of job and silver. Yeah Flash and those were good ideas at a conceptual level, except they were driven by, you know, dominate industry players and that meant that they weren't standards.

They weren't they came as plugins and there was a sort of competitive. Element and just as I think being at open source Summit is a great way to talk about collaboration and community and better together and when we think about the successes of Linux and MySQL and Walzem cloud and lots of great open source projects a webassembly was born and being a standard that everybody can build on and it originally came out of that Skunk Works. That was Mozilla, you know, the creators of rust and they had this vision for basically a simple CPU a little tiny virtual machine that you can put in anything and this little idea has grown into this incredible community and is now supported across all major browsers all major chipsets.

So it was another idea that may have been designed for the web. But our belief is that the impact across the edge and servers and you know server side development will far exceed what you can do just inside the web because now that we have a universal CPU that runs the same everywhere we have the ability to Right software that can truly run anywhere and this is an open standard with participation by organizations everywhere. So webassemble.

It's only the fourth language for the web. And just to be clear, but I think back to your point webassembly is governed by W3 at this point. Yeah.

Yeah, it's it's there's a governing body that's called The bytecode Alliance that's the sort of Steward leader of the community. They pull together organizations such as cosmonic Microsoft and lots of other orgs pulled together to sort of strategize. However, the standard just like HTML CSS and JavaScript is run by the w3c and that gives Chrome and Safari and Firefox and Edge the ability to have a common building point for building this and this is not you know, science fiction.

This is fact. This is here today. This was proposed years ago.

And now we live in this world where industry suddenly has a universal open source virtual machine that's already running everywhere. Yeah. I mean, it's really think of it like an OS it's we're starting right at the beginning.

Right? And Adobe is a big big user of webassembly. So all of our web-based creative products and acrobat and you know Photoshop that kind of stuff that if Have that web experience that's using webassembly.

Got it. So. It's all great.

rubber meets the road and I want our audience to kind of really wrap their head around this now William, you mentioned a bunch of organizations. you know that a cause using with wisdom, but why some excuse me, but Explain to the audience now. Where where does Robert what what are the applications they're using with this where it really?

Look, look, let's talk a little wrap if you will. Well, let's talk about webassembly in a couple different ways because first it's the technology that you don't need a strategy to adopt right because it's become so ubiquitous already. For example, if you're using any of adobe's incredible products that they've moved to the edge for performance for security for ease of use you're already using webassembly.

If you're using things like Google Earth or if you're using design tools like figma, all of those amazing experiences are built on this and webassembly's even being embedded into other software platforms. Yeah. So a big example Amazon Prime video right that they there's a great blog post and that's why Amazon's actually part of the bytecode alliance.

I believe one of the reasons they're part of it, but when they update their apps on embed it, you know, it's really an embedded play right? It's well, we're gonna we're gonna take all our C++ code and we're going to make webassembly module and that's going to be the unit of one at the units of deployment really really fascinating stuff. I love it.

It's why it's such a big it's so hard to talk about webassembly. From a from a you know starting at Ground Zero because it's it's just it's so huge right? It's the prime the prime article was fascinating because they talked about the challenge of supporting over 8,000 unique devices and CPU combinations.

So think like your Samsung television your iPad, you know, all the whole plethora of diversity. I give you another company epics. I don't know if you're familiar with epic Channel whatever it's a channel but it's streaming service.

It's a lot of things and I had this conversation with them they because they play on everything in anything actually some of it through Prime actually, but here if yeah, exactly if you're a company and you're looking at this Edge today, you know, you're saying wow, how do I support this incredible array of devices? How do I get as much reuse out of my code and webassembly becomes this sort of standard deployment mechanism now, I think your question was as well. If we have this new, you know, I think each age of computing has been defined by some Market leading.

Or experience in your question was well. What's the killer app this this question? I'd actually maybe turn it around a little bit.

If you look at the breadth and diversity of who's adopting webassembly, the killer app for the cloud was probably AWS. It was a platform that you could use to build your own experience. Yeah, and I think that the killer app of webassembly will be a platform that enables other organizations to build their own dreams and visions and that's what I hope we've done with Blossom clutter open source, right, you know, we launched this.

All right now, we're the onion. Yeah, come on, you know that we got to see. All right good.

When we started this Vision, we saw an opportunity with with webassembly and our Market leading position comes from the fact that we had incredible management. That's all the value and the necessity for something that's going to be that transcendental and that important to be open source because it needs shared governance. It needs community.

And that was why as soon as we hit, you know, it doesn't developers. We put some Cloud into the cncf and from there it is expl. Did to over 120 core contributors alone.

We have people like Intel and BMW contributing tensorflow and Onyx a Microsoft Onyx which is another machine learning framework models to the community. And that's something that when it's put in there folks like Adobe could pick it up and use it for example, and so that to the question of what is our killer app for webassembly. I think it's going to be defined by the platforms that everyone uses to build your own experience and I think with a real difference between AWS which was ultimately with you know, virtual machines, right all the all the paths are all the infrastructure service.

It's now it's a real Challenge and it's a challenge for these for the big vendors because it's what is is it functions of services? Is it like Lambda or is it like Lambda Edge is it like is it like, you know ECS right where they have where you kind of give them a container and they run it so it's like well, it's really all that right and where it runs almost doesn't matter anymore. And so it's very hard because they've set up these silos and these companies.

Oh you're You're the edge people, right? Oh, you're the compute people. It's like well, it cuts across all of that.

So it's a real I think it's a real challenge moving forward for the big the big vendors. Well, and I think there's two really important facts that your audience should learn when we when we think about that calm. Thank you.

I think that was a very important. The first is is that a lot of the silos that we build in today Are Walled Gardens and a Lambda and Lambda are incredible products and they're built by an awesome company with the team, but it's still gardening. There's no you can't take that.

You can't take that across that distributed Edge or into your own Fleet or your own devices. So again, we're back to better together and community and the second opportunity here. That is a little bit more nuanced is is that because now we're talking about Shifting the cost of compute across that distributed Edge this technology is going to open the market to entirely new business models when you no longer have to pay for compute.

You can have the compute spin up in a user's browser and have them run the microservices there. You suddenly enable new. A peer businesses or even new Hub and spoke businesses that are you just you change the model change the whole different parts model, right?

I mean because one of the things that I think enabled Cloud adoption wasn't just yes, it was virtual machines. Yes. It was AWS that made it.

Kind of I don't want to say easy because God knows it's not but but that made it palatable for everyone to swallow, right? But also there was a course final as a matter of fact, you look at the whole finops thing. That was here Monday, right?

You see how big that was because this cost model is spiraled out of control. Most people don't even have a handle on it. Yeah, right.

I think that is an indicator species that there's a problem. Of course it is, you know, when you have a whole cottage industry developing around it's right. It's required.

It's a huge if you're if you're in the the business of cloud today, you are also in the business of you know, grooming your Cloud cost and managing your guard. But that screams to me we need something better. We need a better match drop here.

Exactly. Well and I think now doctors great and it's gonna be around for a long time, but it allowed that lift and shift like you started with it allowed people to take things that were not made for Docker not made for kubernetes and just kind of throw it in a container. Okay, we're done right and it's incredibly expensive and Incredibly and efficient and not made for it, but webassembly you can't you you cannot do that.

It is it's own OS right? It's tool chain to compile into from all the from these languages and so it's almost it's it's a different kind of Challenge and it's gonna we're gonna see who can do it who can't it's gonna be really sure you know, I I may have a different Viewpoint on this than you guys are a different opinion in my mind almost from the day. Iaas infrastructure service came out.

We would talking about its demise. Everyone was saying this is a fine way point on the way to the real Cloud, which is going to be passed right platform is a service. And we had Heroku and and some really good early, you know tries at this.

I think the fact of the matter is Docker and containers became the past that off grandfather told us not our grandfather, but our uncle told us about right that that became the de facto past built on top of infrastructure as a service and it's great. It served us. I mean, look, there's a whole big industry around this and kubernetes and all that but With with the Advent of the edge and and these other kind of places besides Cloud it may very well be that this translates to that right you can have kubernetes on the edge and all of these things.

but it's a it's a it's kind of like we we Shuffle in the deck a little bit. Right and let's see what comes out this time. The deck's been reshuffled.

And and what I would observe if we take up to the 50,000 foot view is that with each epic of computing we've had dominant vendors. Let's come back to the 90s and think about you know right now and Microsoft and and think about the early rise of the edge or the virtualization with VMware their dominance giving way to AWS the next epic of computing around containers in kubernetes opening the aperture to Microsoft Google and this broader set of and the other Cloud. Yeah.

Yeah around the world but this next epic of computing will also have Titans that rise and tightens the fall it always is and look people still running as 400s IBM still a thing, you know, those people will get to an epic and they still keep that stock price where it is. Yeah, they will be they will get there but Will also see a rise of whole new companies whole new. Absolutely and here's something else based on this new reality.

Yeah, right. So I grew up in the Wintel. You know Empire, right?

It was an evil empire to so I'm not evil to others but it was went out. Yeah, right. The the Advent of virtualization if I would when we first started, let's be of honest right?

We're I don't know how old you are, but him and I have an age when we first started talking about VMS. It wasn't necessarily VMware. There was some other journey I providers that were really good.

Yes, and they were open source hypervises. Well, yeah wasn't even open. Yeah, right.

But it became the dominant player at least, you know. A lot of the public clouds didn't go VMware by default, right? They went open.

But anyway VMware would not probably have been the odds on favorite in the race for which hypervisor rules. Yeah. Amazon web service probably wouldn't have been the odds on favorite for which public Cloud becomes the 800 pound gorilla bookstore all things right a bookstore, right?

He was a Google. Yeah, I mean just but this is why I'm still doing this. Anyway, I think as we look now at webassembly Edge.

And you're right. There are gonna be winners and losers. They'll be some big Winners if this thing really becomes the the New Frontier the new place where we live.

Yeah, you know in the medisphere I guess is the word now, right? They're gonna be winners and as we sit here today we can make bets but if you know, Companies are going to arise applications. They're going to rise usages.

They're going to rise that maybe we haven't even really really thought about yet, right it may be with want some cloud. But man, put it in some. Some person out here's hands.

You're absolutely right and they come up with something that You know, well game-changing look, you know, I I think that I love the analogy of a close company like VMware which is a phenomenal successful business. Absolutely. I didn't not independent again and Amazon which is built on open source and built on.

Yes. It was, you know, they have some awesome proprietary technology. And the reality is the vast majority of the software that you build or you buy is still mostly made up of Open Source and I think that those Trends will continue.

So I think that when we're looking for the seeds of the Revolution we should start with the big open source players and the movements because that's gonna germinate now, maybe picked up by some proprietary player and manifest itself as a pass that's closed, but it's still going to be built together. So I'm gonna disagree with you with all new responsibly. I think the open source train has left the station and to overtake that train you're gonna need some kind of hyper missile or something.

I mean because there's so much momentum. There's so much. Yeah, I mean think of same page then right?

Yeah. I I think we'll be open source. So absolutely they're in Thomson.

Yeah. No I cut myself this morning and I'm worried that I'm bleeding but it's okay. I don't think it's okay.

So anyway, but I think that's I mean all the so webs, you know wasn't Cloud. That's definitely the leader here. There's a lot of companies that have you know that are oh that entered the space every single one of them first thing they do an open source project you have to today it is the fact that business model it builds better software, right if you want when we think about that this is diversity is our strength when you think about the reasons why open source builds better software it's because you have more opinions more perspectives.

It's tested against broader use cases. It's transparent. So, you know, everybody knows where the bodies are buried, you know, you know where things are strong and where things are weak.

And it gives you the ability to collaborate and I think what's awesome about you know, we're at a essentially a Linux Foundation event here, which is a foundation machine that creates incredible properties such as the cloud native Computing foundation with organism part what open source with transparent community and governance really is the accelerant that wins and you look at kubernetes you look at istio wasm Cloud you look at all of these projects. There is a broad theme there of collaboration. Let me tell you what else is important.

And this is me, right so do right 30 years in the 30 plus years in the business. I think one of the unspoken. strengths of this model Is the lack of salespeople?

And I I look I'm a salesperson at heart. You're always be closing too. But here's the thing.

I spent the first 15 or 20 years 25 years of my career. Hiring salespeople. I paid him a lot of money.

Yeah, and they worked like hell to sell. technology today I know if I would hire, you know, quote unquote sales people if I'm selling Technology Solutions. Yeah, because today and the whole the real power or one of the real Powers behind the open source Revolution is get the software into people's hands let them use it delivered Delight show them it works.

They'll buy it. Yeah, I don't need some and I harking you and also the nice thing is that what webassembly in the browsers are already a big thing. It's already an established.

Thanks. So they're just gonna say well, why can't I run this in the data center, too. I made this great webassembly module one right now.

Baked in and that is that is one of the important things right? You don't have to mothball your your data center, you know, right? It's part of this you want to call it a mess.

You want to call it a web. You want to whatever you you're infrastructure. Yeah, right.

It's kind of bored. Like nope. No pun intended to our friends at Google and Oregon kubernetes.

But but it is kind, you know, we will assimilate you absolutely. Yeah. Well look, I think you're you're motion on the bottoms up go to market, you know developer to developer.

You know, let's just emulator in our bombers here. There's three people that matter and software choices. It's developers developers development, right?

But that's the world today it is it is and it's a Persona that's building and communicating software to themselves to other other developer personas. And that's even the way that Colin and I'm that I mean I had no idea that adobe was even using wazam Cloud until we get this, you know inbound email to Hop On Hop on a call and you blew my mind in the first call with what you As we're doing internally it was incredible and the whole team was just like wow, that's awesome. And we've had that that same experience with our community to talk about the Strategic choices for us to start open to start building together with BMW, you know as another example BMW just came to kubecon EU and they talked about how they're putting wasm Cloud into their cars early in production and running machine learning models that can transcend boundaries, you know that can run across their proprietary openstack or data centers or kubernetes.

They're Cloud instances, but also directly in the vehicles themselves and that's a real power that continues to build better models there. I get it guys. It may not he's not good.

We're over 30 minutes for our 15 minute interview. But so I've got to rock wrap up button wrapping up. I'd like to bring this up.

com. com where you spell that for sure. We thought why be stuck in the clouds when you come to space, right?

Okay. com. I love it.

Adobe you know? Yeah, but we would love to hear more about what adobe's doing with webassembly that that isn't you know, I'm gonna look we're here now, but we do text wrong TV three times a week guys anytime, you know, maybe we need a webassembly regular. I'll talk to you about this.

Yeah, what's up? Anyway, we're gonna wrap up. I hope you enjoyed this.

Extended interview do check out cosmonic wasm clawsimcloud. I think it could really if you know in your career going forward, you're gonna need to know that. All right, we're gonna take a break.

We'll be back with another guest here. We're live in Austin at the open source Summit.