March 28, 2023

Episode 213: Maxim Fateev and Dominik Tornow, CEO & Principal Engineer of Temporal

Maxim Fateev is the founder of, a developer-first, open-source platform that ensures the successful execution of services and applications. He is also the co-creator of Cadence, a workflow orchestration engine. Maxim developed Cadence when he was at Uber, seeing the engineering challenges that come from trying to solve the workflow orchestration problem.

Dominik Tornow is a Principal Engineer at Temporal. He is a dedicated and self-motivated software engineer and architect. He uses conceptual and formal models to make complex concepts tangible and complex systems correct by construction.

Julian: Hey everyone. Thankyou so much for joining the Behind Company Lines podcast. Today we have MaximFateev and Dominik Tornow, CEO and principal engineer of Temporal a developerfirst open source platform that ensures the successful execution of services andapplications. It's a special treat today to have both you on the show, not onlyto discuss, what you're working on at , but obviously your background, yourexperiences, kind of the open source space as we were talking about before theshow.

That's so exciting in terms of itsability to not only. Be, available technology for other developers to use, butreally kind of perpetuate the cycle of improving technology and fosteringcommunity. So, we'll get into the weeds of what that's like, but just for, justto kick off here Maxim, tell us what you were doing before you started thecompany.

Maxim: I was an engineer. Iwas always IC an individual contributor in large companies. I, I came toAmerica to in 98 and I worked for a startup, but. I switched to Amazon and Ispent eight. Spent eight and a half years at Amazon. Mm-hmm. . And then I alsoworked a couple years at Microsoft, three years at Google, and last four yearsbefore starting the company at Uber.

So I was always just an engineer andlike tech lead for projects. I never managed a single person before startingthe company.  

Julian: Oh, really? That's,that's incredible to think about that transition and what that means towards,not only building technology, but also creating a product and a team around it.

And Dominik, tell us what you were doingbefore you joined the company.  

Dominik: I had a similarbackground to Max. So, I was at sap. I came to the United States in 2008. I wasat SAP for about 10 years. And then I joined Cisco for about one year. Iimagined myself to be at Cisco for many more years to come.

But then yeah, I crossed Temporal's Pathand I was absolutely fascinated by the technology itself, and I was very, Bythe founding team and the team. So then I made the jump from the largeenterprise to the startup world. And like Max, I. Never managed a singleindividual in my life, but unlike Max, I still don't manage a single

Julian: It's so incredible tohear that you both come from large companies and, and transitioning into the startupworld. I'll start with you, Max. What Maxim, what, what in particular did yousee that that larger companies are doing right in terms of their operations orinfrastructure? And what type of maturity did you bring from that experienceinto tempo when you started building the company?

And what were some quick lessons youlearned when starting to build teams? And, and grow. Once you, once you'vekicked off the, the, the company.

Maxim: I think the, the mainthing large companies have, they have infrastructure, right? Infrastructure foreverything. Infrastructure for hiring, infrastructure for engineering,infrastructure for like marketing and so on and so on.

And at startup will you practically needto first learn all these things from scratch and then execute on them. Likeobviously hiring probably was the hardest one in issue. So getting. People andwe started three years ago. Right now the market is very different, but if youcan think about three, three and a half years ago, hiring people and we aretrying to hire, hire locally, initially only in Seattle.

Yeah. And in Seattle, people don't likereally understand startups the way people do it, for example, in SiliconValley, right. Or San Francisco. So for us, and we also type of engineers, we wanted,we have engineers who could get job in any like top tier company. Yeah. Andcompeting with those companies and gaton engineers was actually the hardestpart probably initially.

Julian: Yeah. Yeah. It'sincredible to think about, that, that, especially when you started the company,how rich the talent was and, and, but also the startup landscape was still kindof, in, in centralized, in certain locations like Silicon Valley, New York,Miami or with some Atlanta, but now, Completely dispersed and across the boardand it's so fascinating to see the growth of that industry.

Dominik, a question for you in regardsto the technology. What were some of the, the problems that you saw and, andhow did you get involved with some pro? Was it at, your current company and youstarted using the product? Or did you kind of see it in an open source formsand start working with the technology?

What was that introduction like and whatparticularly got you excited about what they were doing at Temporal?  

Dominik: I came across Temporalthe first time during the pandemic, and I'm sure everybody had a similarexperience during the pandemic. You found ways to entertain yourself. Yeah. SoI had an ever-growing reading list.

Yeah. And eventually I came acrossTemporal, the te. And I was like, okay, that's interesting. So throw it on thereading list. And then I made my way through the reading list and then I thinka month or two months later, I came back to it, to that particular entry. Andthen I was like, okay, let's dive into this.

Yeah. And then I looked at thetechnology and Also at the implementation. And when it clicked, I wasabsolutely floored. I was absolutely floored. And in that moment, well, notquite in that moment, but actually the very same week. While I was stillstudying up on Temporal, Temporal reached out to me and then I was like, okay,I have to get on this call.

And then talking to the founders andtalking to the other engineers on the team and I was. 100% sold. Yeah. Becauseas Max said, right, so large companies, they have the infrastructure and oneaspect of that infrastructure is they have huge teams, right? Yeah. Sap, ahundred thousand colleagues.

Yeah. Cisco, I believe around 75,000colleagues. But then at, at Temporal, at that time, I was of course massivelydownsized. 20 potential colleagues. Yeah. The concentration of talent in these20 people was just absolutely mind blowing. Yeah. So the combination oftechnology and, and people and yeah, that was, that was an easy decision.


Julian: Yeah. It's incredibleto think about, especially it's so in, in startups it's so critical that earlyteam, and it, it's, it's that collaboration, it's that ability to, one foundertalks about like, it's the type of people who see something on the ground and.Versus just walking past it and ignoring it's that kind of proactiveness and,and mentality that really drives the success of a lot of startup.

Maxim a question for you in regards toopen source and this whole concept. I think people are becoming more and moreaware of it. It's becoming more and more of a household kind of concept. Butdescribe was the, was Temporal always open source? Did you, did you ship it outinto kind of the world to see if developers would use it?

What's the story behind kind of thatprocess and, and collaborating? All the developers in the community thatstarted to use the technology. And, and when did you know that it was viableenough to, to build a company around and start, start, start using that servicein, in other companies across and, and that they would need it? What's thatstory?  

Maxim: I think if you wantkind of story, probably you start a little bit earlier, as you mentioned. I, I,I spent eight and a half years at Amazon. And actually I was there twice. Soit's totaled eight and a half years. And I joined Amazon first time in 2002when it was pre relatively small company.

It, it already was public, but it wasaround 800 developers. Certainly not comparable to any like, what it is rightnow. And back then Amazon was the first company which decided to go all in onmicroservices, because when I joined it was actually one big monolith. Yeah.The problem is that monolith was that practically you could actually run thewhole Amazon website on a single machine develop machine, which was kind ofcool.

You could put break points and play withthat. But problem was that it took 18 hours to compile, like mainline on a itpowerful machine. It's not even developing desktop. Yeah. And then every timeyou change something relink in that thing, what was 40 minutes? So Brian, 40minutes relink in that applic.

So sometimes you can do dynamiclibraries and like, and avoid really like, slow linking, but still developmentprocess was very, very hard. And they realized the only way to kind of scale isgo to microservices. We call them microservices these days. Back then it wasservices. And I kind of witnessed the whole thing and I was take lead for thepractically infrastructure component, which was project pops up, like themessaging on Amazon.

Mm-hmm. And it was well before Kafexisted, right? So they practically built our own stack and I was take lead forthat later Simple queue service. When Amazon AWS appeared like web services andSimple Queue Service took that as a backend. So that I think that engine still isstill. So, and we kind learn that Amazon, that this type of like hu andmessaging is a cool idea to kind of coordinate between microservices, but atthe same time we realized that it's actually pretty bad way to do microserviceorchestration and or do like a coordination between, because it just becomesvery complicated.

Yeah. And we started to think aboutorchestration solution and we had free internal versions and the public versionof that was AW pub simple workflow. Aws, S W F I was the lead for that. And we,the first time introduced ideas, which we right now have into portal. And wedidn't kind of completely nailed it down, but at least these ideas were played,played.

We played with those ideas. And thenlater practically when we went at the tub with microphone, micro founderSummer, who also worked with me on the simple workflow, we build messagingsystem. And is an open source. And but later we kind of switched to this othersystem. We called it Cadence which was kind based on the ideas of the simpleworkflow service from Amazon.

Yeah. And we, well, at Uber, we startedto do it as open source from the beginning. Yeah. Uber allows to do that. Andwe practically put it out and we were developed. We can go there right now inGitHub and see all the chickens While it was. And the interesting part is thatwe drove adoption with an Uber.

Within three years we bought up to ahundred use cases, but first two years we didn't get any external adoption. Itwas cricket. People just were, okay, it's GitHub, not that many stars. Gothere, check it out. Nothing happens. Right? But then two years, then for somereason we started to get adoption and first companies were like for example,one of the first who started to use us and then Airbnb, Coinbase.

So we actually got pretty interesting,like digital native. Which embraced the technology. And then from now on, Ithink we ended up on the hacker news. Yeah. And people knew I went to a fewconferences and then it started kind of to pick up and open source communitystarted to grow. Now we have around 55, I think hundred people in our selectchannel, but we believe many more developers use us, like probably aroundhundred K people already in using our technology in some shape or form.

Yeah, and actually just to summarizethat, we didn't build it from scratch then. Like, as a company it was always westarted a Uber ran for four. And then we realized that the only way to reallymake it popular and like a life scale kind of open source project is to have acompany behind it.

Because we could, we should focus on anexternal adoption on people in the world who want to use that.  

Julian: Yeah. Yeah. And realquick before I go, you, Dominik follow up question Maxim is, is what is, whatis the process like to getting a company to adopt, say, an open source tool ortechnology just for those who don't know in the market?

We think about open source. Some of theindividuals who are less technical, like myself, we think about open source asreadily available, usable, and then kind of like you can kind of, yourengineering team can use the SDKs to, to, start building more efficiently, moreeffectively in all these tools.

But what is that external adoptiondriven by, in terms of a company standpoint? How do you, how do you identifywhat targets to focus on? Who's adopting it, what companies you want to getthis technology into, and also how to. Something that is available online.  

Maxim: So from our companypoint of view, you mean? First most of our adoption so far was organic. It'snot like we went to companies and like we certainly presented conferences. Wedo go do talks, we do blog posts. Yeah, but most of the adoption actually camefrom initial, from Mugger. Cause Uber was large company and Tur became verypopular. As I said, when we left it was around hundred use cases.

Now it's probably thousands anddevelopers move around. Right. And in the US at least, developers changecompanies very frequently. Yeah. And we, every week somebody comes and says,interpret, I use this technology now I want to use it. My new. And that is, wasthe major I think. Just people talking to each other and bringing to othercompanies.

That was the major, I think, adoption.So, and as it is also open source is like very different types of open source.One thing about Temporal, we have very permissive license. We have MIT licenseand we have Apache two for some components. Yeah, so practically, which meansthat most companies have no problem using open source with this license.

Obviously if you have some, like someother more complex. Company can help some, like other companies can say, no,no, no, no. Like, let me go for lawyers. Right. But if it's mit, mostcompanies, okay, you can use it. You don't really need to prove anything becausethis is a very permissive license. So a lot of our adoption happens Bottoms up.

Yeah. People do poc, people dohackathon. Just put it in one use case. Yeah. And later way our software worksis that, It's decay library, as you mentioned, that you write your code and soon. And there is backend component, more like database like service, and thatwe only monetize by just running the service for you.

So you experience how you develop andhow you run your code, how you deploy your code. Doesn't change. Don't thing.It changes. It's like has host database in the cloud, right? Yeah. You can runyour own database. You can use hosted one. The same thing hosted  

Julian: Yeah, yeah. And m  

Maxim: details if you wannalike about monetization stuff.

Julian: Yeah, . Yeah, no, forsure. I'm curious, Dominik, from, from your perspective as, as an engineer andas a builder you, you've worked with different types of companies and I'mcurious from a, a building standpoint, what has been some of the benefits to buildingand op with open source kind of tool and technology and availability, but alsowhat are some of the challenges that people may not know?

Kind of the source code is available andpeople are, able to use it. What, what are some things you like about it? Whatare some challenges that come with building in, in, in the way you arenow?  

Dominik: So I wanna argue thatwhen I entered the ecosystem, right, there was, there was this I would sayturning point.

So you could say, when I entered theecosystem, we just crossed to chasm from. Like open source Norma to open sourcenative. Yeah. And then surely within from 2010, 2000 to 2020, the open sourceecosystem completely exploded. Yeah. So I did see the, I did see the change,yeah. That, that this explosion brought with it and the most, Powerful impactto me was that even so SAP and Cisco were large companies.

They were closed in the sense ofcollaboration. Right? Yeah. Yet once you have an open source project, there isbasically no boundary. There's no limit. Yeah. The open source project itself,The ecosystem defines its community. It is not defined by a company anymore.Yeah. And with that you can Work with incredible talent Yeah.

Across company boundaries. And that isfor sure this in hugely empowering and hugely uplifting Yeah. To be workingwith people who are just like passionate and interested in that particulartechnology, in that particular project and who are not confined by company.Yeah. So I have to say I have only good things to say because I am very muchan.

I'm very content and technologyoriented, not I, I just, the business problems are not on my mind. Sure. They,they're on Max's mind, not on my mind. So I have nothing bad to say. Sure.About an open, open source and the ecosystem that form around it.  

Julian: Yeah. Follow andfollow up question, Dominik, for, for the engineers in, in the audience.

Care to kind of reach principal statusand, and that can mean many things. It could be the first at a company, itcould be the most senior level engineer, but really it's the responsibility andthe ownership and all that that kind of encompasses and, and I'm sure a lot ofproactive communication on your end, thinking about ways of technology could bebetter.

What are some advice that you would giveother engineers out there who are either the principal engineer current? Or, orstriving to be one at a company that they're passionate about. What are some ofthe best practices that you have for accomplishing the work that you need to,and, and what are some things to think about if you are an engineer?

Kind of up-leveling your career, workingon technology and thinking about ways to improve it and collaborating with thethe founding team.  

Dominik: So, in general, butthat is true for every every stage in your, in your career is never stoplearning, right? Yeah. Always be curious. Never stop learning.

Yeah. That is fairly generic advice. Andmaybe border us on a, on a cliche, but that is for sure. Got me to where am Itoday? That is that curiosity and that learning opportunity take that learningopportunity. Yeah. And then of course the, when you, when you cross the chasmfrom more junior to more senior, you take more ownership and you take moreinitiative.

And that can come in many, many forms.Yeah. So for example, at Temporal, what I very much enjoy to do is to talk toour ecosystem. So, I am a lot out there on the, for example, Temporal slack,but also conferences and so on, and meet with people in our ecosystem, meetwith our users, whether they are customers or not.

Customers, meet with our users and thentalk about their use cases, talk about their requirements, and then work with them.On their architecture. Yeah. And that is personally there's like, that is anactivity that I hugely enjoy. And at Temporal I do have the possibility toactually engage in that a lot.

Yeah. But there is any there, there'sbasically lots of different possibilities, right? Right. How you can up levellots of different dimensions. That's one of them that I chose.  

Julian: Yeah, that'sincredible to think about. And, and also, and, and thinking about more aboutthe, the company and, and the technology and the business problems we'resolving.

Maxim, I'll, I'll shift this question toyou. Are there any use cases you've seen that are particularly exciting thatyou didn't expect Temporal to be used for the companies or industries? How farreaching is the, the technology and, and how many companies are benefiting?What is the overall benefit, if you were to sum it up for companies who adopttechnology, whether it's sophisticated or simplified as Temporal?

Maxim: Yeah. I think weprobably wanna, at some point talk about what actual is, right? Yeah. , becausecertainly it's, it's, it's, it's easy to say, we, you don't get benefitswithout explaining what it is. Yeah. But if you just talk about type of usecase, we are very practically if you think about it, we are target fordevelopers.

So we are, our only product is used bydevelopers directly, which helps to develop a. . And so developers need towrite code and write their application using our approach. Yeah. Which makestheir life much easier because they need to write much less code and their codegets reliability on infrastructure level.

Yeah. And we can go into the details ofthat. They, because it's so generic, especially new way to Right, distributedreliable systems, it's applicable almost everywhere. So practically every timeyou need to say, I need something. Well, I need to make sure that my data isnot lost and my, my kind of business process or whatever keeps, keeps going inthe presence of all failures to is a good fit.

Yeah. And almost everything fits there.If you think about it, infrastructure automation, deploying data centers,configuring clusters , a bunch of APIs there lifecycle of almost anything likefrom databases to like, big flin cluster. That is one. For example, hasor, oneof the first use cases, they practically built control plane for their cloud.

Like Hasor didn't have cloud offering.Then they started to build cloud offering. They needed an engine and they usedTemporal for that. Ah, that is one. But then you go up, you, you have forexample, things like big data or ML pipelines. You need to automate. And it'snot only just you need to practical automation around that control plane aroundthat life cycle of those, that is where we use the a lot.

And then you go up level, you can dopayment system for example like a, a lot of FinTech and banks use it for to dopayment. Yeah. Or and practically ensure that you do payments. And Dominik, youcan like give, for example, just talk about how you do like a saga, sagapattern using our like how it fits our approach.

And then you go up and you practicallydo business level. You do things like or DoorDash, or you do practicallybusiness. and we've seen and then you go like IOT devices and the managing iotdevices and so on and so on, and practically almost everything when you say, Ineed this to be reliable via good fit.

Yeah. And that's why like the number ofuse cases is practically explored and it's almost everything. Yeah. And it'svery hard to say. Case doesn't fit. It's very few.  

Julian: Yeah. And describe alittle bit more, so I know you alluded to it a little bit more in talking aboutthe use cases, but describe the technology and the mechanics behind it, just togive our audience a breath of, of what it does, how it works, and, and if I'm adeveloper, what should be my expectation for the experience I'm having?

Maxim: Yeah, I think I'll letDominik talk on high level. He's, he's good at that.  

Dominik: Good . So actually Maxalready got started on this. So, Max characterized Temporal. As transactionsfor distributed systems and I very much like that analogy. If you think aboutdatabases and transactions, transactions have great desirable properties,right?

Yeah. And that property is usuallysummarized in the acronym as ethnicity, right? Consistency, isolation, anddurability. , but that super slick acronym shadows a little bit of veryfundamental property of transaction and that is the guarantee of completeness.Yeah. When I start a transaction, I get the guarantee that the transactionexecutes effectively once or it doesn't execute at all.

Mm-hmm. . With this exactly once or notat oil semantic, it is almost trivial to transition the system from aconsistent state to a consistent state. Mm-hmm. . Mm-hmm. . And now Temporalactually does the very same thing. Not to databases, but to distributedsystems. Yeah. Temporal offers a core abstraction, like a transaction that wecall durable execut.

and the durable execution is really justa function execution, but with strong execution guarantees, yeah. That functioncannot fail and that function cannot time out. So as a developer, you get anamazing developer experience because you can literally write code. As ifplatform level failure and platform level timeouts, they don't even exist.

Yeah, right. We have very impressiveexamples where we implement a monthly subscription workflow where once a monthyou charge a credit card as a for loop, a never ending for loop. Yeah. Thatsays charge this credit card and now sleep for. Yeah. And this function doesn'teven like account for the possibility of like a crash fault or a timeout,because Temporal makes sure that for that function, that possibility does notexist just like a transaction does.

Yeah. And with Temporal workflow, a lotof problems become trivial because the underlying cause of the problem. Isalready taken care of.  

Julian: Yeah. Wow. It'sincredible to think about the way the system is built and, and how honestly atpeace a lot of developers can be with this, with this continuous operatingsystem.

And it's amazing to see the use casesand the, the, the amount of traction you've seen in the market. Justorganically, and I'll shift this question to you, Max. What are some of thebiggest challenges that Temporal faces?

Maxim: I think the mainchallenge I wouldn't say it's cha let's say it is, I wouldn't say it's likechallenge, challenge, but this is something that which kind of slows down us isjust awareness. I'm pretty sure there are a lot of, of your listeners whichnever thought about Temporal. And what we know is once people learn about it,they get pretty excited.

Yeah. And this is very useful platformand we had some limitations throughout languages but now we support all themajor languages. We had recently added support for type, script and platform.Yeah. So every developer out there, but we also support Java Go agent supportfor net and later Ruby. So we are practically every major language right now.

And so. Every developer can use it as,as, as, as today. And I think it's mostly about awareness. That's why we aretalking to you, right? Because we wanna just, your, you can, you don't need totrust me. Just go to our website and it's open source. You can goto documentation, to tutorials, education, and join our Slack community andmake your own opinion.

Yeah. But reality is, as Dominik said,like at some point, like bulb will  

go on and they, oh, right. Thistechnology kind of will save me a lot of time and effort and make a kind ofcompletely, I will need to completely rethink how I build systems. I, we see itall the time. But if you think about it, the rest is execution.

A nice thing about our startup, ourexperience wasn't. Standard for startups because how part of the startupfighting this product market fit, right? Yeah. Like this magical product marketfit. In our case, we had product market fit before we started the companybecause we already had adoption, external adoption of the pro open sourceproject, and within Amazon and Uber, and also Microsoft, because my co-founderdid Siemens and Microsoft.

Yeah. Ever heard about durablefunctions. Mm-hmm. , he was the author of the Durable Task. So, we kind havethese three big companies where this thing became pretty popular, so we kindaalready knew that this idea will fly. It just for us it's mostly execution,like startups should execute and growing company building, company hiringpeople, doing all these other things.

Yeah. So not simple, but this isexecution, how to execute.  

Julian: Yeah. Yeah. Dominik,I'm interested to hear your theory. I know you, you were gonna speak on it, butplease,  

Dominik: Yes, I have a, I havea, a theory and that is, that is an ecosystem we are not used to identify andthen embrace new core abstractions.

And that is because that actuallydoesn't happen very often. So one of the first core obstructions was operatingsystems, for example, a. Right before I had the core obstruction of process. Iwas responsible to manage the CPU on an application level, right? But thenprocesses came around and the process gave me the illusion that my myapplication owns the CPU all by itself.

and similar happens to memory andvirtual memory, right? Yeah. Yeah. I gave the illusion that memory isunlimited. Yeah. Then a new core abstraction few years later came around thatis database transactions. Right. Database transactions gave you the illusionthat this transaction that cannot fail and it runs in perfect.

I. Right. Yeah. But then for a very longtime, no new core abstractions came around. Right? And I think that tempo, thedurable execution is one of these core abstractions, something that lets youthink differently. . But since we are all process natives and oil transactionnatives, we don't really think about it.

We just take it for granted. Right. Butwe are durable execution nomads that transac that core abstraction is stillfairly new. Yeah. But we don't have a lot of experience how we internalizethat. Yeah. So I do believe that me is spot on. And he says one of the corechallenges is basically the, the hearts and minds.

Of the developers and communicating thatcore abstraction.  

Julian: Yeah. Yeah. And I'mcurious to hear from both of you, if everything goes well, what's the long-termvision for Temporal? And I'll start with you Max. What's the long-term visionfor Temporal?  

Maxim: Well, obviously we canmove company in a lot of different directions, but I think the minimum, likelong-term vision is I think the most important thing is that this open sourceproject to succeed independently from the company, right?

Like Tempo is a open source project. Ithink the, at least I want every developer know about it. Yeah. And ideallyunderstand what it provides. And then obviously it doesn't fit a hundredpercent of use cases, but at least in use case where it's a good fit, I expectit to be used. The word. Yeah. And we also, I believe this time there'll bemore products doing, using this global execution approach because again, it'slike, like transactions.

There was one database probablyinitially, and now we have a lot of them. Yeah. But then longer term as acompany, I believe that we can build company, which will simplify life ofdevelopers because we can just manage a lot of complexity. Yeah, and mostlyaround how to run that. So every developer will be able to just take sdk linkthe application, start using that, and just point to their, our backendservice.

And we will take care of all the hardproblems for them about making reliable scale and so on. And I'm not eventalking about complex things like multi-region deployments, revolvers, and. Andwe have a lot of those ideas and actually things in production as well. Yeah.So, lot of things we'll take of, and I see potential.

This market is practically immensebecause every company has thousands and thousands of use cases, which in thismodel.  

Julian: Yeah. So  

Maxim: it's certainly not abig market. This also makes our VCs very exciting.

Julian: I can imagine. AndDominik, would you add anything to that? Anything that, that you think in termsof the long term vision for the technology and the company and, and where thingswill.

Dominik: So yeah, the companystrategy that all rests on Max's shoulders. But given the technology and thecore obstruction, I believe that durable executions they will grow into thecloud programming model. Mm-hmm. and so, and up to this point, right, we havefigured out how we provide.

All right. With the very same onecommand line command, internal command, I can provision one machine on aws, 10machines, a hundred, a thousand. Yeah. There is no limit. Yeah. But runningapplications. Right. Is a very different ballgame if I run it on one machine,on 10 machines. On a hundred or a thousand.

Yeah. But then when we have somethinglike durable executions as a foundation of a cloud programming model, then thebetween running your application on one machine, 10 machines, a hundredmachines, or a thousand machines is is actually like, just disappears. Yeah.And so in short, I wanna say for Temporal, there's basically, there's.

Only one way. And that is durableexecutions everywhere.  

Julian: Yeah. Yeah. I lovethat. And I know we're coming close to the end of the show, so I wanna makesure I ask this question cause I love how people abs, essentially digest orabstract knowledge from things that they ingest. Whether it was early in yourcareer or now, and I'll start with you Max and then I'll go to you Dominik.

What books or people have influenced youthe most?  

Maxim: Books of people. I,okay. I read all, all, all the kind of standard books back then, right? Likeabout patterns, about whatever, about architectures and so on. Yeah. But Ithink the most important thing was just working with people. And I was superlucky to work at Amazon.

Yeah. And I think Amazon especially like,like it just got very high concentration of talent, especially in theinfrastructure space and services. And just witnessing like beautiful thingslike s3, like just say the best and just understanding how people think, howthey approach problems. That was, I think the biggest one is just just it's noteven one person.

It's just I think Amazon has just such ahuge concentration of talent. Yeah. And types of problems that we are trying tosolve. Solve, and the approach to those problems, like trying to solve for thewhole world. It's extremely hard to build service, which you expect can kind ofthe whole world run on and.

S3, I remember they did presentationinternally. When uh, they had, when they planned it, they had a certain kindof, okay, what would be the most, most scalable kind of solution we need tobuild? And they had, okay, let's build whatever for this scale.  

And then they showed like a few yearslater that they built the scale, like multiple orders of magnitudes.

So you, they couldn't even imagine thathow big it'll be. Yeah. So I think that witnessing that in person, I think wasthe kind of, for me, the most something which influenced big time, how I thinkabout systems, how we design systems, how we design to, cause again, if you wantevery company to, to kind of rely on us.

This service should be some, notsomething you, you need to think from beginning, from first principles. Youcannot one thing you cannot do, you cannot add scale later. Yeah. The idea thatyou will build something and then like magically edit later just doesn't work.This is one thing about simple, which is cool, is white startups like us forexample, a bunch of white startups use us because we speed up development bigtime.

Sometimes we hear five x improvements oflike speed of develop. But at the same time we work with, even if you have likevery small scale, it, it helps you to develop faster. Yeah. But because webuild it for uber size can business, and we run a scale from the beginning.Yeah. We this, we know, this technology can practically grow even, even more.

Well practically your startup can growis splash. For example Snapchat. Every Snapchat story update goes from. And we,and there are a lot of snap snaps taken during for the new z and so you knowthat you can go with Snapchat size, but you can be small startup and you needto change your code.

It just keeps running and scalingmathematically for you.  

Julian: Ah, it's incredible.And, and Dominik question going to you. What books do people have influencedyou the most?  

Dominik: So, one of the earlyinfluences was in college. My professor, professor, Dr. Vent, Siegfried vent,and he is a father of the fundamental modeling concepts.

And he instilled the notion, Of systemsthinking. Yeah. Into me and conceptual modeling. And that was very early on.Got me on the systems engineering path. And then the, and the next turningpoint was discovering Leslie Lampert's work on t a plus the Temporal logic ofactions.

And that turned out to. Not just amethod or a tool that is a thinking tool. Mm-hmm. , it shaped my entirethinking around systems and distributed systems and was hugely valuable. Wasvery much an up-leveling game. So, Of course, I'm obviously biased, but Iencourage everybody to look into conceptual modeling, into formal modeling andinto systems modeling.

And do get started with the works on TLAPlus because There are also other authors besides Leslie Lambert that use saidmethodology and the sync thinking tool to teach others how to approach, yeah.Complex systems.  

Julian: Yeah. Yeah. It's beensuch a pleasure having both of you on the show and last little bit.

I always like to ask this questionbefore we share your plugs, your websites and everything like that, where youraudience can find you. But I'd love to ask you both. Is there any question thatI didn't ask that you would've liked to answer or that I should have asked you?Anything come to.  

Maxim: I think one thingwhich I want to just mention is that we have a developer conference inSeptember.

Yeah. So we had a, our first conferencelast year and it was super cool crowd from and you can find it online on ourwebsite, but I absolutely recommend you to join our conference in September.Even if you don't know new to Temporal or you already know a lot about that,so. We have members of community who have with us for already six years, but atthe same time we, we have a lot of, last time we had a lot of people who justcame first time.

Yeah. Like you said, we saw it somewhereand we just came to conference to learn about it and I think that was prettyamazing. So please come to our conference. You don't find the references, so nowebsite and everywhere.  

Julian: Amazing. Yeah.Anything I missed on your end?  

Dominik: So I just also wannaoffer some more material.

And that is last year in 2000 22. Ispoke at the Strange Loop conference about Temporal. And the title of that talkis Workflows New Core Abstraction. For distributed systems. So for anybodywho's interested, especially in like the conceptual foundations the theory thatpowers Temporal, then I wanna recommend that a strange loop talk that is onlineon YouTube at the. Strange Loop Channel.

Julian: Fantastic. Andobviously we'll share this with our audience. As we have a huge developercommunity that we, we are in constant contact with from a company I run AutoLabs and also the different founders out there and different CTOs that arebuilding and, and looking for better and more efficient ways to build and, andespecially as they scale their, they're startups, like you said, it not onlyworks with startups, but for larger companies.

So we're obviously gonna share this withour communities. Our developers and our founders, but let us know where we canfind you, where we can find you on LinkedIn, where we can find Temporal give usyour websites, give us your LinkedIns, give us everything that we need to knowto come support you two but also to support the company and what we're, whatyou're working on.

Maxim: So, yeah, my onLinkedIn is just my last first and last name. @mfateev is my Twitter. But go toour website and You can join our community Slackand we have support forum and we have, and on Slack, you can find me there all thetime. So it's not that hard to contact me directly if you.

Amazing. So I would recommend to Gorolike, to kind of be in in, in connection with our like project and mepersonally.  

Julian: Absolutely. Dominik,yourself  

Dominik: as I hang out on Twittera lot @dominiktornow. And I tweet a lot about Temporal of course, but also alot about distributed systems in general and databases and transactionprocessing systems.

And if you want to talk about Tempora;,yeah, if you want to talk about like your use cases, your requirements, andthen check out how tempo may fit in your existing landscape, or how tempo mayfit for your next project, then do find me on Just on theintroduce yourself a water cooler channel or send me a dm directly.

I'm super happy to get together on zoomor any other platform. And then we, we talk about your project.  

Julian: Amazing. Max andDominik, it's been such a pleasure chatting with you. Not only learning aboutthe story behind Temporal, but what you're working on, the use cases that youcan connect to and how people can get in touch.

I'm so excited to. This technology growand ultimately, all the technologies that are similar to yours in the way thatthey just help companies run more efficiently, be better in terms of thetechnology and, and the, and, and enabling their developers to do more workand, and spend less time on redundant tasks or tasks that, don't, don'tnecessarily need to take the time that they, they do.

But anyways, thank you so much for beingon the show. I hope you enjoyed yourself and I hope, I hope you enjoyed beingon Behind. Company Lines today.  

Maxim: Thank you.  

Julian: Of course.  

Dominik: Thank you for havingus.  

Julian: Yeah.

Other interesting podcasts