jamie allen interview revised
Recording: jamie allen interview revised
hello everybody I'm Alexa crabber of the organizer of SF Scala and here we are on location trade shift where we have an awesome meet up about microservices we have to talk today in one of the taxes from light band and here the last we have Jamie Allen he is the senior director of global services at light band which until recently was known as type safe and obviously it's everybody's curious so maybe I'll just jump right into it and I'll ask Jamie you know while I bend what does like the mean okay well one of the things that we found about our name aside from the fact that people would typically call us typeface mm-hmm was that it didn't convey anything to the executive stakeholders that we were trying to market to when you're a company that's trying to build a business around selling technology to executives it has to be a name that means something to them and whenever we would go to a conference like Gartner group they have a something in Las Vegas called a Adi I don't even know what it stands for but we'd stand there in our booth and people would walk by and look at us and say type-safe and then they'd walk away mm-hmm they thought that we made something like tool to help them type better or fonts and that's not good but more importantly one of the things we realized about her name was that it bounded us it said that we believe this and while we are firm believers the type safety is critical in buildings you know correct software who knows what the future holds so we didn't want to to put a boundary around ourselves and say that this is all we do and really if you look at the technologies we produce now not all of them are fully type safe so this isn't a great descriptive name of what we want to be and so we thought about how we could rebrand ourselves in a way that was non constrictive and that's where light Bend came from the idea that it's an aspirational name and it sounds like nonsense but it really does have a meaning to us and that we're trying to create technology that bring the world to the way we think and whether or not we're right we'll see but we have ideals about how software should be built and we think that we're a gravitational force in the world to kind of bring people to our way of thinking mmhmm yeah I think it's kind of you know since the time it was introduced I kind of found that it kind of grows on me right and and and it's actually it's a very good explanation I think it's kind of interesting because you know type safety is one of the things scholar community believes it's very important you know another's functional programming all right and but I think that the the overarching belief is that we deeply care about software engineering we have experience based knowledge how what the best practices are right and want to bring people to to to to work with us using this best practice okay so this interesting way to put this wrestle you kind of you emphasize that that gravitational pull right right so so so to tell a little bit the particular i know that you can of your one of the guys that like band who most kind of one like most meetings with with actual customers you you kind of you know you you see what people are doing right you know how skull is used in practice so you actually do a lot for actual promotion of our kind scholar use in in in in industry yeah and and so I wonder right how that influences your thinking what do you see what kind of trance can you just tell us a little bit you know about kind of your career in the industry and how how do you see trends how you operate yeah absolutely I think historically I've been with type safelite been now for four years actually on April first yeah and this auspicious day congratulations thank you the we have seen a complete transformation and the kind of companies that we work with now there are the early adopters out there the LinkedIn's the the apple and a lot of them are technology focused and then there are people in the financial sector we see Morgan Stanley is a really big proponent of Scala you see Goldman Sachs being one of the companies sponsoring Scala center mm-hmm so finance companies are very quick to find technologies that help them solve problems and they adopt them quickly but in the enterprise for the larger companies that are traditionally a little less risk willing mm-hmm right they have started coming to us in droves and now it's a matter of trying to help them all in a way that is meaningful and it's really tough it's actually changing the way that we do business one of the things that I'm going to be taking on more and more is a role of strategic enablement where companies who are saying we want to go with this technology all the way we want everything to be that everything that we do going forward to be with this kind of approach mm-hmm and whenever I go in there I talk to people I don't even preface it with the whole hears us here's what we do it's more the philosophies the the ideals that we're trying to get people to believe in and once they see that and they grasp them and they embrace them then we see them you know want to go this this path all the way and that's going to be my goal going forward is helping companies you know be successful holistically not just with the proof of concept or a single one off or anything like that will be much more about strategic enablement mm-hmm so who is the audience are usually when you come into a company who do doctors is the kind of senior architect chief architect is a trip of engineering is the CTO so who is the decision maker on the other end why historically it's always been we've had a champion somebody who really believed in what we were doing was bringing us into that company mm-hmm and then there are certain companies who see this you know and in very segmented companies who don't work as a entire team you know one group will adopt something another group will adopt something and you know that's been very scattershot for us but now with the people that we're talking to it's been much more executive stakeholders the CIO levels mm-hmm and whenever I say this I mean for the largest companies on the Internet as far as enterprise goes I mean you can talk about Amazon and stuff like that and they do use our stuff mm-hmm but whenever I'm talking about say big hotel chains right retailers companies that do a lot of Internet of Things because it creates so many devices like GE or Samsung we're certainly talking to the people with the highest levels of those companies IBM is an excellent example of this iBM has fully bought into our platform and we're now training hundreds of their people mm-hmm and we're building all these videos for them they're such big proponents of spark and they want people to use spark from Scala mm-hmm so this is super exciting for us yeah that's I think you know IBM kind of joining spark ecosystem and kind of realizing Scala benefit through it I think it's really huge for the community right because like we previously seen IBM back in Java and back in Linux right so it's put spark in the same league exactly it's a third big push they've done of a strategic nature and if you look at what they're doing it's it's really impressive because they're not just saying we're believers in spark they're investing in it they have a team of people here in San Francisco working at the spark TC yeah and their entire job is a hardon spark yes you know build out the documentation write the tests make sure that all the comments are in the code and also contribute real fixes that make this more powerful so when you have an entity like IBM behind you technology can't help but be successful as far as you know being marketed out to the enterprise community and it lowers that risk barrier for adoption yeah so so of the kind of the one of the core messages of types a flight band is reactive systems reactive platform yeah and can it took a little bit for those folks who are not familiar with this terminology what is reactive why is it important in the interpreter prize and basically how are you aligning this new vision for the company to basically reactive to to the world right well we started out with the ideals that we wanted people to embrace and that was the you know the concepts of being responsive to whatever clients you have whatever whether their users or b2b or you know whenever you're receiving information from fitbit's and that all is driven by being able to scale up and down being elastic to all the different events that can occur inside of your world but also being resilient and not losing things just because something went wrong at the hardware level were you know all those things that you really can't even type check and we feel that underpinning everything to make that successful is a message driven approach it's very easy to get our head around this developers because I don't know very many people who have encoded against a message queue like a you know mq series from IBM activemq or RabbitMQ or even think in terms of Kafka mm-hmm right which is more widespread message boss use the thing in startups yeah it's certainly certainly in the starts but also in the enterprise as well I mean huge numbers of enterprises are using Kafka now this is a great way to buffer all the load that's coming in to you before you consume it mm-hmm and you know that's that's that allows you to think in terms of well given that it's in some sort of queue it doesn't matter who the consumer is anybody any server any service can can grab that data and do work with it and that enables the elasticity that enables the resilience and and that leads to the responsive experience that we want so that's where we started mm-hmm and now that people have embraced those notions we're moving more into this idea of okay how do we do it fast mm-hmm how do we take that data and really quickly make decisions and we see this with customers and some really unique spaces like online gaming and by that I mean gambling or you know whenever you're placing a bet William Hill in the UK they have a tremendous business around this where they're you know they have to immediately make decisions to be profitable based upon the odds that they're offering people who are placing a bet on a horse or a game or whatever mm-hmm but also in the financial space because that has to happen in real time you get it okay the less profile yeah so this is how people make money and so Dean wampler has been really instrumental in helping define this canonical architecture of how we believe system should be written to handle fast data while still meeting these reactive principles mm hmm interesting and so indie was also driving the big data agenda for live band right and she is a champion of spark so however how do you see spark and reactive inter interact I see them as sort of orthogonal concerns mm-hmm and that spark is merely a through which you get access to information where you have all this data coming at you've got to make sense of it so the whole idea is you spark to perform any kind of anomaly detection or just supervise models where you have a train model that knows that based upon historical data I can categorize it a certain way and therefore make assumptions that it will hold true in so many percentage of a case like ninety percent of time this will hold true and therefore make business decisions in real time automatically as opposed to having people try and figure that out on the fly based on charge state see mm-hmm so so it's so I'm wondering how do you kind of see the light band strategy in the enterprise so so you unveiled the legume microservices platform which you can describe today in the technical talk yes ah so so but obviously kind of and that jump first right so and now so you kind of you know you obviously were the biggest fiction the scholar pond yeah and you will be pretty medium-sized fish in the Java pods yes and the everyone you start right so and but you've been in this industry for man year so you know the javis i don't think so i wonder how do you see this idea that kind of my advanced gonna compete in the bigger javeon world basing on its skull advantage on that vanish acha encapsulate yeah how do you see this strategy unfolding in the enterprise yeah that's a that's a good question what we what I think is speaking historically we have at our company you know enterprise customers mostly right and these are companies who are trying to move into the space at a time when there's a big paradigm shift for their developers it's a lot for people to ingest and try and think about asynchrony non-blocking concurrency parallel ISM you know forking and joining and how I'm going to build you know it's like the graphs of all this information and stuff like that it's it's difficult right and so a lot of them make the choice to say I want to start with Java because I don't want to throw a new language in there as well mm-hmm now at life and we're obviously huge proponents of Scotland we think that is definitely a step forward in application development using a language hmm but that said we can't ignore this market and historically we have if you look at the conferences that are available for people using play framework or acha they're all in Scala if you look at even spark usage I mean how does somebody using Java find out information about spark and this is an area where we have to make up ground for the rest of the world mm-hmm we have these customers they need to be successful we can't ignore them so they have to be first-class citizens yes and I think one of the things about our announcement of light Bend and legume that was difficult for the Scala community to embrace was here we came out guns blazing for the Java side of the house mm-hmm but we'd ignored them for so long I mean it was all one hundred percent Scala yes we're you know fifty percent of our users are Java that's right there's a big disparity between what we're doing to help people as opposed to what our actual customer profile looks like so we have to help those people there will be a conference this year in austin in october focused around using these technologies from java mm-hmm so that there is a voice and also a place where people can send their teams to learn yes and that's going to be really important that said we think that Scala is a linchpin for innovation we see this with all the great frameworks that are coming out now you do not see this kind of innovation happening with Java or other languages you know people were doing more with big data using Scala than they are with any other language and this includes Python I mean there's incredible libraries available in Python for doing computation yes but that doesn't mean the engines exists for doing this at at scale correct at the core level it's supposed to let the user level exactly and I think you're right in terms of big data space we know we did the Big Data skull conference last August and it was doing you know my cousin was there materia was the G Krebs was there Martin himself keynote at with a message that basically spark is a scala gsl yeah which i think is signifying the development ah you know when I was to kind of thinking about this actually thought that there is something you can teach the skull community and I think this live band of Elton is really great and urs kind of interfacing with businesses also you know can really teach scala businesses a lot right because essentially and I want to run this by you now I wonder what your thoughts are so so you to work for amazon and at amazon what struck me is that every venir thinks it's a business owner engineers very businesslike because they make decisions essentially they want to satisfy the customers like so if you see that customers order is stuck in like in the queue you are proactively doing whatever it takes to unblock this customer order right then kind of so you know without any regard for hierarchy or whatever so so so if engineer things is a business owner great things happen in Amazon kind of is evidence of this right so so you guys really found a niche where you can take the scholar goodness wrap it into java api put it in front of businesses see adoption yes and see you know growing market share so so i think this is actually good for for skull like a system in general right as because it lets you sell more scholar by services it right it lets you have most collar jobs and build moscow systems if they expose api so kind of you know ninety percent of the world is using already right your GV I'm already so I wonder you know if this is actually a a good model for basically software engineer shops using skull because you can basically you don't need to be like band to do this right you can be you can be any kind of consultancy who happened to have smart engineers who believe in the way the architect systems but since you already are on GV m and you can expose Java API so you can go after businesses where the traditionally have been Java shops right so so do you think this strategy is transferable and kind of you know can you guys teach the rest of Scala world let them in on the fun right like I let them all make more money with this I think that's true and there's an opportunity for the Java side of the world as well we see this already actually in spark with the data frames API where you see the type level people building this incredible new API dsl for type safety because data frames use a lot of string lee typed arguments we're just passing in strings that is you cannot be type checking at runtime you find whether you mistyped something so that's totally unacceptable as far as API design goes for people who care about soundness in correctness whenever they're building the system that has to go into production they don't want to find out three hours after they put something out there that's wrong so you see the Scala community already moving to build an API for data frames this I'm safe yeah and I think somebody you know who's really crafty could be doing the exact same thing for java mm-hmm it's not that there isn't type safety and Java it may not be as stringent as what you can achieve in Scala but that doesn't mean that there isn't value right and if you look at legume because we came out with it with Java first I mean a lot of that is also very practical mm-hmm it's a lesson learned from having acquired play framework an acquired spray framework that it was difficult for us to then take those frameworks and just move them to a job API you couldn't just bolt something on yes and so for log on in particular we really want to make sure that we built a fluid Java API and then we can do amazing things with Scala for Scala API but we had to make sure Java was right first right so that was a really critical aspect for us cool so that's I think this is very interesting and I mean play was available in Java right already so like one yes so I one that's right so so you guys kind of in in a good position to struggle this both worlds and and so in terms of microservices obviously so everybody's mind right and it's just occurred to me I mean you know I and I've lived to see and do some korba and I just thought in a wood is you know didn't call by invent everything that market services are renovating now you know what is kind of what is crucial which is different this time around right we had kind of many many goals at this thing and and so in how and obviously you know our case is very powerful piece of software and and also you guys are coming out the goal and they also have conductor yea which was not discussed a lot but I'm really impressed by what it can do so can you talk a little bit using it at night or when we looked at it right so but it's kind of its its new right so so so how can you kind of what is your vision for my care services from light band side how acha a conductor played together and like what is the differentiation right right what is the key difference here which gonna can be the goodness LED band brings to this space right I think two things right first of all one thing that I think has historically been completely ignored by framework builders on the JVM is developer experience mm-hmm right it's incredibly hard to build just a simple RESTful API using the traditional tools of Jersey and you know whatever frameworks you want to use you gotta patch them all together and there's just it's it's awful to build up a war and then I have to figure out are you gonna hot deploy or are using Jay rebel or something like that you know maybe you don't maybe take the server down replace the war bring it back up all that kind of garbage hmm and that slows people down on let's face it's not fun mm-hmm so one of the critical things for us was play framework is a really great developer experience when you're building web apps yes right you've got the ability to sit there and in your browser just hit f5 refresh the browser and you see the compiler errors right and you don't have to worry about anything about whether the code is pulled into the server or anything like that it's all taken care of for you mm-hmm we wanted to bring that exact same experience to people who were developing restful api is that don't have a front end mm-hmm right they don't have some sort of user interface that you see hmm and while you're coding you save it's automatically redeploying it for you in your development environment so you can test whether or not this is going to work really easily with a built-in Cassandra backend for your persistent entities and then this is where the second part comes in we're really focused on what we think is the correct way to build these services we do not want people building them with crud based monolithic data stores that are then going to be shared among multiple services we think isolation is the key not the size of the service microservice is a terrible name we want people to be focused on isolation you should not expose anything about your service beyond the API the people call and that should impose no dependencies on them that they have to follow to make sure that they're compliant with your API and that's why you see rest being such a success because people are able to just make HTTP calls right the only real concerns are things such as service discovery and gateways and and versioning and stuff like that but those are more implementation details than they are whether or not you are able to decouple these services yes you want your teams to be able to operate independently you cannot have shared libraries between them you do not want to have anything like you till or miscellaneous or Core as a you know shared library of dependencies it just represents a mishmash of garbage that all gets pulled in by all these different services because and this is what Ben Christensen was talking about on microservices common it's not mm-hmm trying to make sure that you're going to be able to upgrade something like guava across all these services whenever you have shared libraries is it's a mess mm-hmm don't do that right don't share a database every every service should have its own data store with its own schema it is not shared by anybody else the only way to access that data is through the service hmm these kind of ideals or how we think people will build scalable software and and by scalable I don't just mean being able to handle all these requests but scalable Crossley teams okay I ganization look exactly and that's our goal that's where we want to take people and we'll see if we're right it's very opinionated approach and log on you can turn these things off mm-hmm but we're gonna make it hard on you we really want you to follow this model and this is this is on the front end there's gonna be a lot of companies out there would say no and I don't want to do that MSA fine go use spring boot I don't care mm-hmm but we owe sum of all that yeah you know whatever don't use it it's fine it's a great tool as long as you like servlet based interactions and you know crud based databases well these are things that we think are anti-patterns especially at scale you want transactions you're doing it wrong how do you share the API definitions can at least share they can ask to shed the API definition between the services of their only simpatia lutely and i'll show that today okay where I can bundle up using SBT mind you this is one of the tricky things for us right we don't want to be a company building a build tool yes but the things we can do with SBT that allow us to wire things together such as the developer experience in logo in play is so powerful that we we can't move off of it we had really heated discussions about what we're gonna do for maven wanted to support may even as much as possible and we simply couldn't the way we want to so with SVT we can sit there and wire together a bunch of services single deployment footprint and it sounds like I'm saying monomyth but it's only a monolith from the perspective of my build file yeah it's not the deployment itself it's all going out there independently using tools like conductor yes and then you're able to you know say run all and everything just runs and it exposes it through maybe you gotta play framework or spring mood or whatever restful api that you can then call and then everything behind is using streaming asynchronous api's as opposed to you know this request response kind of stuff you don't want that everywhere in terms of my everything I've just been at the top level northeast scholar conference so there is this new tool called cbt which chris just wrong every moment and it's kind of funny to see in occupations that here here old man in 30 lines of scholarly she does know ninety-five percent over it so hopefully maybe next year in all the community kind of around matures and you know there'll be an interesting thing but that's a great thing about scholar we don't we have a lot of different things and community can kind of help and Chris studied at the EPFL he was part of the typesafe team as an intern for a while working on slick yep so you know we have a great relationship with them and we're absolutely excited to see what he does with cbt I mean this is all part of having a vibrant ecosystem and you know like i said if people want to use greater we're all for that whatever that doesn't bother us right we want people to be successful that's what's important building software that works cool so I I'm very excited to bother the new developments or if I think this is gonna be the type dish is gonna lift all the bolts yeah and so we're looking forward to your tech talk and thank you very much for coming you bet thanks