In episode 24 of the Big Data Beard podcast live from Strata Data San Jose, the team dives into the world of batch and stream processing with guests Fabian Huesk from Data Artisans and Flavio Junqueira from Dell EMC.  Our guests detail an open source streaming data stack consisting of Pravega (stream storage) and Apache Flink (computation on streams) that offers an unprecedented way of handling “everything as a stream”.
Title: Changing the Processing Paradigm with Pravega and Apache Flink
Host: Cory Minton
Co-Host: Robert Hout
Guest: Fabian Huesk & Flavio Junqueira
Links:
If you enjoyed Strata Data San Jose or were not able to attend, don’t worry, Strata Data London is right around the corner and we are giving away a full conference pass!  Check out the blog that details how you could enter to win.  As will all Strata Data and AI conferences, use promo code PCBEARD. to get 20% off conference passes!

Even 2 years ago will pass or say last year so I guess people are still getting to know where they go.
A before pravega I’ve done a bunch of things affect I was doing research before before.
At what is a Yahoo search than Microsoft research and then I work for counseling for a for sometime and then I joined another provocative mean and IM see that’s what I’m doing today I’ve been doing open source for quite a bit of time.
As I was I was one of people who started projects like Apache zookeeper in and Apache bookkeeper so I’ll be involved with Apache projects for for quite some time and I find it very excited to work an open house projects that we like.
If you’re working with the community and inside.
Cory Minton:
[2:22] Very cool so you guys are doing a taco today cuz I find it interesting that you guys are here together and you’re going to tell him that I joined story but your your title later today is as title Unified.
Inelastic batch and stream processing with pravega and apache-flink so help me understand why you guys like what’s the story with these together and then let’s dig into it.
Fabian Hueske:
[2:44] Yeah so did the story that we are here telling us so we’re basically Sharon division of yeah it’s a sad unified Bisons game processing.
We see these today these concerts are basically seen as to.
Distinct distinct processing prayer times and in our vision we want to share that you can actually do that with the combination of two systems.
But integrate nicely with each other and Men or talk we going to talk about how Rebecca and Flint.
Can be such a combination to achieve this unification affection screen process.
Cory Minton:
[3:19] Okay so for the folks who don’t know so batch processing.
Is pretty common in the another Hadoop land rights on that produce obviously be in the early sort of batch processing many of the other Frameworks that have come along since I’ve been sort of batch oriented is that accurate,
so then what’s the like why is streaming so important what is streaming do for us that we can’t do with Bash.
Flavio Junqueira:
[3:44] What’s the one thing is processing data any real time as soon just in time trying to get insights and you know very quickly over the day that they are in Jackson.
So you can think of a number of applications that can benefit from this if you’re trying you’re trying to get recommendations out of a you know whatever set of events that that is coming in it’s important that you process those events in a in an orderly manner.
Right because if you want to determine that a things you know one.
Get one type of customers is purchasing one thing and then purchasing the next one and you want to do this or recommendation to another customer you’d like to know that.
So those kinds of a of a that kind of processing and dick out of a recommendation that something you can do with your new real time.
Crossing over the date of the other spaces other demands of applications that are also interesting.
People talk a lot of a lot about iot this stays right so sensors sensors producing.
[4:44] Lots of samples of a time and in are you processing all those samples and trying to make sense of all that data sometimes it reacting to the interchurch it up servation to make out of there are the sensors connected cars connected Home Center and all that stuff.
Cory Minton:
[4:58] Sabaton screaming and in many many large organizations can have is if these Technologies.
Can I came into fruition a lot of folks adopted this concept of the land architecture where you had this sort of streaming you know capability that was for that near-real-time sort of analytics and then you had this batch that was.
You know yet you still had to store all the data in some persistent file system but then you created these views that were logical based on the needs of the business right and that was the concept of Landon if I if I know if I’ve heard her correctly about some things if link is after another’s is.
At a table made of a slightly modified architecture I think the folks called the kappas at the right.
So some of the terms so help me understand what you guys are doing is it is this a and that this is part of the land architecture where you’re continuing to just do that differently less lossy can a more accurate or you.
Truly like changing the overall data pipeline in architecture.
Fabian Hueske:
[5:51] Answer the main motivation for the land architecture was to provide Frost the results for for analytics dashboard.
You would have like a man in the classical Pax Brothers in science what collect data for certain amount of time then if your batch processing trip to Crunch the Raider and right at 8 hours to some some Cafe Restaurant later or let her use it.
The land architecture you would have added real-time system that processes the day to answer this as it arrives but at that time and its architecture was proposed the systems that screaming systems when that mature.
And only could only produce a proximate result so they results better not not exactly the same as batch process I would have produced.
Modern stream technology stream processing technology you can actually get much closer to do this trip to the results that match president.
Cory Minton:
[6:50] Okay so cuz that’s one thing so stringent when it first started out that the concern was that.
If I want to go fast I get real-time then I had to be willing to give up Fidelity right accuracy which scares people like in a recommendation engine it’s not that big of a deal cuz if you get a recommendation a little bit wrong.
It’s not like detrimental the company but if you were trying to see if you get if you get that data wrong and you make that decision incorrectly like somebody’s life is at stake.
So is what you’re saying that these new architectures are allowing us to not have to make those selections between speed and fidelity.
Flavio Junqueira:
[7:24] But it’s smaller than that right side only Fidelity you also have to consider that if you if you do the lungs architecture you have to datapads to maintain.
Straight sew-in if you have you have you found in operations you have to deal with both by Plies The Real Time one which is perhaps not accurate but that doesn’t have high fidelity.
And the other one that is suing batch-processing.
Rob Hout:
[7:48] 2 digit today to buy blinds to code sets to prove everything right so it’s the main take the maintenance of that kind of environment probably comes down.
Flavio Junqueira:
[7:56] He can be up yeah it can be a nightmare.
Cory Minton:
[8:01] Help me understand what because Flink is a project that I know is is gained popularity over the last few years and it’s super cool to talk somebody else sent but part of the team that’s bringing that to its Russian tell me why.
Flint came to be like what its history was and how it kind of fits in the in the ecosystem.
Fabian Hueske:
[8:18] Yep so blink before I started out as an academic research project so.
We initially designed that to be something like a hybrid system of meth use and relational database systems.
I forgot database research group so we packed a lot of database principles in the India processing engine we had an Optimizer to optimize certain the.
[8:43] Processing strategies right from the start and we had a pipeline execution engine such as.
And such are the ones that out today the best systems have and when my disease design decisions pretty much probably on when we brought them Flink.
Gave blank into the edges of a foundation.
We noticed that I’m still processing was really picking up and we were looking at bass guitar technology stack of the technology that thing provides and if they got well actually it’s not that far away from what copas has to offer and.
We are basically at all these features that are necessary for a modern screen process such as.
Events in processing handling of lot State large state trooper from Statesboro complications of a streams and evolve these pictures now for A4 quiet quite some time.
Cory Minton:
[9:39] Okay so what is the so if somebody’s looking at you know there’s more to get in the Stream processing like what’s the big Advantage at a kind of at a higher level on what Flink does that maybe.
I don’t know what storm or even Kafka the might do like what’s the the big pitch on why I should be really digging into flank.
Fabian Hueske:
[9:58] So think I’m coming to the full spectrum of all streaming use cases.
It’s handled such a sad it was designed as a batch process that also have not found it streams would be cool like that. It’s reprocessing.
Very well as well as a screaming analytics use cases but also have Android applications like the kind like the evolution of microservices architecture.
Hannah’s all these all day the photo full spectrum of batch processing.
Cory Minton:
[10:27] Okay so then if I wanted to play flank is it something that’s,
is it shipping with you know what your standard distribution of of a Hadoop distribution like pick your favorite one with his horn works or Cloudera or the open-source versions is fling shipping is part of that package or is it something that.
Lives underneath the Zookeeper or yarn heard in terms of service how does it actually get deployed.
Fabian Hueske:
[10:48] So if you think it’s not included in any any hair while it’s included in elastic mapreduce Amazon.
[10:57] It’s not not included by the other Defenders.
Flint hamady price on yarn so you separate can run it as a yarn service it also runs on messes with a new release that will come out in a few in a few weeks I will also have a prime support when it is so you can actually.
Keep LifeLink pretty much everywhere and it also runs as a standalone system if you if you don’t want to set up a resource manager.
Cory Minton:
[11:21] Virtual so so we’ve got to link which handles Downstream processing very cool why do we need for Vega what what’s the what’s the reason why pravega exist and why is it complementary to what you’re trying to accomplish.
Flavio Junqueira:
[11:36] Right there are there two main reasons for pravega to exist.
When is any change s data and typically you the one couple the source that is producing data to the stream processor which is Lincoln dinner in the in the in the discussion.
But that’s something you don’t want to do it don’t do that coupling and so you need some some component to ingest that data.
Traditionally this has been done with messaging systems are you look at a right angle Kafka author of messaging but substyle substrates that its substrate that that ingested it for you.
[12:15] Problem is those systems I’ve never be have not being developed.
They have not been designed and developed to host a top for snowfall for long periods of time for extended periods of time so if you have a string that is long granny has me running for and I know maybe ears.
What you have to do is you have your recent past or than those systems and then your distant past is stored somewhere else movie to hdfs maybe and then going to do any processing you have to go to that other sister.
What causes your complications to in passing that data because you can easily think that out when you want to do this over the data you might want to combine,
you know that that the recent past with the with the distant past due joints over that and it that that kind of processing and if they are across separate systems you just complicates the life of webassign.
To do that and so pravega is a system that allows you to ingest data and just keep it there in whatever form you choose.
Cory Minton:
[13:12] So like when you say whatever form like is it does it go and live in a file system does it go live in an object store like our visits until it with writing data.
Flavio Junqueira:
[13:19] Obstruction is a string so you write you your pantry stream you read from a swing by say it nearly guarantee order.
A pretty bases that that kind of stuff but so we got into that ordered that is given by the application and the and the any process the data in extreme format.
That’s one of the things that I that we’re trying to achieve overall is going with protected by the overall vision is that applications.
Are the process data in string format.
And how you need to Acacia of baton stream is is along those lines right so let’s not differentiate batch and stream everything can be processed as a stream.
Rob Hout:
[14:00] Find a storage nerd it so I have a question what what what what makes up what makes when it comes to actually lay that down and we like what do I write so it may be calling it a stream but I don’t know what’s in.
Flavio Junqueira:
[14:16] So abstractly for the application what he means is that we are storing the events you’re giving us in order.
Cory Minton:
[14:24] It’s like an immutable log file right.
Flavio Junqueira:
[14:27] Pretty much an internally while we do is to some extent expose that as well we have we have segments.
[14:36] And the segments can be either open or seal.
[14:40] And we use those to compose streams right so you can you can you know you can think of a stream as a as for example of a concatenation of a sealed sidemen in real Consignment and when you went to the open sign.
[14:54] Overtime you can see you. Just like when you create another one and with that you can you can stay that I can spread that I download across across a number of service and I’m not limited by the capacity of a of one single machine.
But there are cool things I can do with that right with the satisfaction of segments another thing I can do is skate with string.
[15:16] But because I have one single segment if it’s hot it’s taking too much load I can see all that great you and have just given you power Listen by doing that.
Cory Minton:
[15:25] I’m running across multiple machines that are.
Flavio Junqueira:
[15:27] I run across multiple machines exactly right and in those two I can split into four and I don’t know so depending on the workload I can we can I just.
Capacity for Gaston forestream Basin on that one we observe.
Cory Minton:
[15:42] Knows I can find to a to a Cluster meeting like in a location or can you actually segments that scale over distances between like data centers between clouds.
Flavio Junqueira:
[15:54] It’s okay instructor yes we haven’t done that work yet.
Cory Minton:
[15:57] Okay. That’s totally fair that’s right.
I’m so so that’s that’s interesting so what’s the.
What’s the path for for consumption of afflink and pravega like how to people meet us out to dinner since I’ll be so you guys bring Flink to Market and what is your value are you providing services are you providing licenses like what’s your your delivery model.
Fabian Hueske:
[16:18] Yeah so there are dozens provides distribution so which obviously includes open source link but also.
A proprietary component which we call the application manager and does application manager basically maintains the life cycle of applications of streaming applications these applications typically run 24/7 so they need like a special.
38 Special Care when you want to maintain them and something to drink or price of each other Frank provides are safe points while you take consistent Snapchat the whole state of an application.
Put that in hdfs find some more S3 and then you can restore the application from that from that site once all the source I reset it to that position of the safe Point all the internet’s verification reset.
And this is a very powerful tool because you can do things like stop and resume.
An application you can also run a pee test you run different applications starting from the same safe but you can migrate an application to another data center on Euclid version and all these things nicely integrated in this application manager.
Cory Minton:
[17:25] That’s incredible that’s pretty interesting cuz that’s one of the things I think.
A lot of the the modern applications of struggled with which is that traditional it thing of I need to be able to take a point in time snap of a application recreated for Natalie purposive disaster recovery.
How about I take a copy of that put an hdfs set a replication job so that if I have a smoking whole disaster I can actually rebuild that critical application that’s that’s pretty interesting.
Fabian Hueske:
[17:52] If I may interrupt you so we’re actually have a user who’s using this feature for an awesome.
You scared so they are basically checking safe points and always going for the for the originator of yes that has the lowest spot into so they’re running their applications on spot instances and whenever something fan stitches migration to the next to the next week.
Cory Minton:
[18:12] See that’s pretty slick.
Think that that’s what I see I was trying to get back that’s what I want to know is like what’s the use case I forget that you’re building locations in this is a framework that allows you to do that but it’s not that’s cool I could actually solve the problem of decrease latency improve performance that’s kind of cool.
Rob Hout:
[18:27] 1 calls.
Cory Minton:
[18:27] Yeah I know exactly what are some of the other like that it I saw there’s a ton on your under the date our sins website you got great logos of customers that are using it what are some of the unique problems that they’re solving by partnering with The Artisans.
Fabian Hueske:
[18:40] Yeah so are the friends i n g is using using fan for a fall protection.
I like using that too young for to detect fraud and in financial transactions.
Yeah I thought you just are using flank to to build in House services based on secret screaming secret or.
[19:04] We also know of a company that built the back end of her social network using Flink so that this is like an event-driven architecture design pattern and every click on a website basically.
[19:18] Create an event that goes into into into a m s h q stand deliver to flank handled by Frank and tragus also.
Other operations and again this company uses light side points for for princess for AP testing 2 to test different recommendation.
Cory Minton:
[19:36] Shoulder like if it’s on just my mind starts going on anytime I hear like event processing already I start thinking about like.
You know machine-generated data logs like so thinking about log analytics and is there a is there a used case or do you see a future where that could be used for large-scale log information and analytics for.
Things like preventive maintenance if you think about the iot constructive dealing with how do I make sure that.
Assets in the field are being consistently monitored so that I can analyze are virtues when we take them out of a so you know how to production is that kind of a future that kind of a different use case.
Fabian Hueske:
[20:08] No I think that actually fits the the division of think that pretty well like out of this place.
Cory Minton:
[20:14] Very cool Southern pravega obviously is an open-source thing but I think it’s been announced in in previous conversations that.
You know there’s some Dell EMC project going on there where they’re going to have this project code name which one of what Starbucks is open source today correct is that right.
Flavio Junqueira:
[20:29] Predict is open Saturday yes.
Cory Minton:
[20:31] And then is there work being done inside LMC to do something similar to what they did Artisans has done with having that unique thing that they deliver above and beyond the open-source is that something that that’s being worked on.
Flavio Junqueira:
[20:42] Yes oh so we are we are working a platform.
But that’s not that’s not an official project product yet so he has been announced in some Dell EMC conferences it you know so that the project has been announced but it’s not at IGA is IJ product.
Cory Minton:
[21:00] Very cool but the goal is then to create a like you said a streaming storage platform that allows these stream processors to ingest the beta separated from your traditional.
Flavio Junqueira:
[21:11] Yes on the product side yes but.
Maybe I’d rather talk about more about the community cuz that’s one of the directions that are that that we really want to go so we one of the reasons we wanted we open source early.
Was exactly to be able to attract people and Guetta and Imperial Community around Utica meaning of developers set up perhaps excited about the technology about the cool things that at that we can do with all that attractions that we are proposing.
And that sucks that’s recently that we look into doing more I mean it’s is still we still don’t have many contributions are people interacting but we would discuss One Direction that we want to go.
Cory Minton:
[21:46] Well that’s why you’re here right I’m here talking about it showing the use cases excellence.
Flavio Junqueira:
[21:50] Exactly in talking to you guys as well.
Cory Minton:
[21:51] Yeah I’ve told are you going to go to get a greater than Siri.
So so when you guys when you go prison I mean tell us a little bit about the story like it at a higher level like that what are the three things people are going to hear when they come sit in your session that they’re going to take away and why they should go down with the slides and bring you guys in.
Flavio Junqueira:
[22:08] It’s okay so I’m going to help line to talk and say I guess that’s it that’s a nice way of doing this so we will start by motivating.
Why we are talking about a unified batstream Vibeline.
Price of giving queries along the lines of what I was saying before like you in a so you’re crazy do not need to differentiate between recent passing this and that,
and weather in of the data is is he or if the date is there and how do I make that work right so it’s all coming from the same source.
Vega and I in need to process City Express perhaps in with a sequel with a sequel query.
So that that’s that’s that that’s out of the motivation of of of the work and then a provider provides a number of features that enable the processing with fling.
To be really but you be efficient and effective.
For example one of the really nice things that we can do is is processing with exactly one semantics so in pravega we provide two features that enable that in the flame pipeline maybe Fabian wants to talk about it.
On the flip side by the end of the checkpoint in we do on the reader side and the and the the transaction to provide on the right side and Abel has to have a full pipeline.
I think job reads data from pravega right and Xena does its processing and outstate a bag.
Cory Minton:
[23:35] Very cool did you want to come to the door.
So we just need some in the past and I think you’re kind of letting this earlier like we’ve seen early stages of this and there was I think it was it was it was a dumb SeaWorld last year when they did the the brain waves.
Rob Hout:
[23:51] We did it we did it at at David works.
Cory Minton:
[23:56] So tell me but to be part of that do you know what time at the.
Flavio Junqueira:
[23:58] Yeah I know yet.
Cory Minton:
[23:59] Okay so tell me about that cuz it was I saw people I literally walked by this booth and I saw people wearing these you know look like you know EKG or whatever monitors on their heads.
Well this isn’t being a virtual reality this is freaking me out or what are we doing so tell me about what that like what that wasn’t going to use casewise what that drill.
Flavio Junqueira:
[24:16] Was that was that thing in the head I don’t know that out that’s called to be honest but that was just a bunch of sensors generating.
Events.
I said if application was actually pretty simple that was just some visualization around the events that that add that device is generating I show you know your brain waves and so on based on a remember correct was like a bunch of questions and opinions,
I know how your reaction was then your brain with this play differently so.
Cory Minton:
[24:43] It’s another done that at conferences around the world and you’re not reading Minds that’s what I just took away from that.
The answers to questions that’s very cool so I wanted to give you the chance to talk about your session I’ve certainly we encourage folks to go check out this will the slides be online are you guys going to post your slides online after order.
And I will try to grab the link for those to the folks can check them out weird if I had a folks engage with you to personally if they’re interested in learning more about data artisans in about 4 Vega what’s the best way for them to engage you to learn more.
Fabian Hueske:
[25:17] Yes or no for me the best way would be you can write me on my Apache email address which is probably out there if you want.
Cory Minton:
[25:26] You guys have sales guys that are like out in the field selling for you or.
Fabian Hueske:
[25:29] Oh yeah of course I mean well so far yeah you can also reach out by the day. Is this website to us if you don’t want the approximate.
Cory Minton:
[25:37] That’s good we’ll find you in Apache what about you how’s the best place for how do people if they want to learn more about for vague and they want to learn more about the Dell EMC kind of work on the side in terms of product ization what’s the best ways for folks to engage with you.
Flavio Junqueira:
[25:52] Yes for protecting itself that is a website for Vegas that are you outside our main repository son isn’t get hub.
I’m in fact a starting point to highlight that that’s where we do the work is out there we have an internal branch and push stuff out we will redo the work.
In under you have a repository directly so we’re going to go there going to see all of all the thousand in the same no pushing stuff and interacting a song.
So you just said about the project self go there and I’ll check while we’re doing the things that they’re going on as such if you want to reach out to me directly I mean the number of ways is that is right.
I guess I am not active on Facebook these days but no lie.
Cory Minton:
[26:33] Trash fire you don’t.
Flavio Junqueira:
[26:35] At least LinkedIn Anna and the Intruder II check regularly so free free to connect on those and then I’ll email all his works fpj at a party.
Cory Minton:
[26:47] So then one quick question for you shift gears you’re here this week going to be talking about this this idea of unified in elastic and unified an elastic search for streaming batch.
What’s next what are you guys doing over the next couple months like what are the exciting projects you’re working on our kind of the the direction you’re trying to take the product and give us a little forecast would be great.
Fabian Hueske:
[27:06] Yeah so I’ve been working for the last 2 years on Nexus I’d like the radiation if you guys like screaming Secrets very very hot topic.
I was just asking like being Integrated Health System ski stays.
Blinkist Farragut support for offering streaming Secret Service on streams but also an investigator we do that in a way that you don’t have a special special send text message to Magic so it running the Furious.
Bad semantics with Patterson Tax on streams which also means they can easily part part lyrics from historic attitude originator.
Something that I’m I’m working on into something that I will also continue to be working on.
Cory Minton:
[27:45] Rico becomes at their first I understand correctly for Steven Seagal apis that you can any of those previous queries that you can have built in the batch contacts that you can without having to totally rewrite the way that you interact with stream data sets you can actually take those.
Complex equal language grease and move them into the stream is the day after that.
Fabian Hueske:
[28:04] ESO what what basically happens is that the screws are the end not not processed by.
Investing all day that once crash another day too and debating Resort properties Coricidin incrementally computer.
So I kinda like similar to what battery system is doing when you’re building a materialist View,
so you’re changing the base status of the definition and then these changes need to be propagated into the material rescue so this is basically doing that in, in the context of length that is running late just wait at sister,
High throughput streams.
Cory Minton:
[28:41] Okay so so what’s the next big conference for you are you presenting it any any conferences your next few months.
Fabian Hueske:
[28:48] Oh yeah so they will be looking forward conference in a month here in San Francisco.
[28:55] Yeah so the conference will see quite a lot of interesting use cases of length that would be an awesome show.
Cory Minton:
[29:03] You have like 4 weeks all that on the web and I was like this is cool so how many people are saying that in that have you cuz I know it’s it’s only was last year the first year for point forward.
Fabian Hueske:
[29:11] So we had last year the first fin 14 San Francisco we had it already three times in Berlin yeah so usually about three to three and 50 people.
Cory Minton:
[29:23] That’s awesome and we’ll go over look forward to some news over there maybe we’ll come by and hang out with you fuck it what’s your what’s your next up man what are you working on the next few months Golden Community obviously.
Flavio Junqueira:
[29:34] Yes been community at exit but probably the question is what I’m not working.
[29:40] Thanks a bunch of things that directly doing of course I’m not at the stages same say yes Frank.
If there’s lots of things I still need to do I think one exciting thing that we were really going to be working on his side.
Historical stream processing boosting your historical swing Brasi story in sorry I focus a lot on the on the east side of New York near real-time stream processing story but you know we also care about the distant past data which.
And if we need to boost outside around there to provide a new features on efficiently.
[30:19] So yeah so that’s that’s one thing that we going to be focusing on and in addition to I don’t know stuff like security and all the fun stuff.
Cory Minton:
[30:26] Governance and all the other fun things are you going to be at the flick forward conference as well.
Flavio Junqueira:
[30:30] I’m compared to Free Fall conference then at the data which conference in Berlin then then Estrada London answer.
Cory Minton:
[30:37] Yeah okay very cool you’re making the full conference circuit that’s awesome.
Flavio Junqueira:
[30:41] Community Building.
Cory Minton:
[30:41] That’s right that’s exactly what you’re doing well very cool guys thank you so much for being on it again I encourage folks to check out the the slides will put a link in the show notes for folks into Flink,
today Taurus and what you’re doing there that the unique project things you doing their sounds really interesting and I’ll be super Vega building a community,
if you’re interested in getting involved certainly reach out and get involved in the get Hub in the put it on my work but thanks again for being on guys I want to do a quick.
Shift here we have a little fun section of the show we like to call Rapid Fire and it’s basically our chance to just ask you some fun personal questions,
so here’s that works you basically just sit back relax take a deep breath.
And say the first thing that comes to mind then when I ask you these questions and we’ll just going to I’ll go back and forth so we’ll start with we will start with you Flavia what year do you think Skynet will go online.
Flavio Junqueira:
[31:30] 2013.
Cory Minton:
[31:32] 2030 alright what about you we all are later we’re all going to be killed if it doesn’t okay what’s the last good book you read.
Fabian Hueske:
[31:42] Oh that’s that’s probably brought the book that I’m currently writing.
Cory Minton:
[31:46] Oh you’re writing a book.
Fabian Hueske:
[31:47] Yeah writing about ice cream processing with Apache Flink it’s a still working progress but I hope it will be outside.
Cory Minton:
[31:53] We could be an O’Reilly media hopefully.
Fabian Hueske:
[31:55] Add wavy yeah it’s already in the early early so you can look at it like unedited pets.
Cory Minton:
[32:01] I like it alright will check it out what about you what’s the what’s the best book you’ve read recently.
Flavio Junqueira:
[32:05] I’ve been reading a lot about the political situation Catalunya sorry I don’t know if you guys know about that but I am so that’s it that’s an issue that has been going on so I’ve been writing for myself.
Cory Minton:
[32:14] How good for you stand for and on us next all right so what genre of music are you rocking these days.
Flavio Junqueira:
[32:20] I don’t know why I just go all over the place I go from classic to pop and end in rocket.
Cory Minton:
[32:27] Alright so you’re at your renaissance man with your music what about you.
Fabian Hueske:
[32:30] Yeah pretty much the same like it really depends on the mood.
Cory Minton:
[32:33] Keep it looks alright alright so what piece of technology is currently making your life worse.
Fabian Hueske:
[32:43] That’s that’s a that’s a tough question what is my command I guess like all the all the mobile data situation Germany it’s pretty awful.
Cory Minton:
[32:53] The mobile data situation what’s going on there.
Fabian Hueske:
[32:55] Like all the other providers said it is all very expensive and not very good service.
Cory Minton:
[33:01] Really that’s too bad cuz I actually have.
Fabian Hueske:
[33:03] So I’m my provider Park me here for 2 years are 6 megabytes.
Cory Minton:
[33:11] It’s actually funny I actually bought so we were in Germany this past year and my wife.
Has a phone from the us and we went over there and we ended up instead of buying one in London where it’s an English-speaking country and I could have easily easily navigate to the website,
we waited till we go to Germany and so I got Vodafone Dez on Deutschland.
Like SIM card to put the iPhone oh my God you would not believe how hard it was for me to put money on the stupid thing I was blown away alright so what piece of technology is making your life worse.
Flavio Junqueira:
[33:37] I think that you obvious colors would be printers in video conferencing.
Cory Minton:
[33:41] That’s good that I like that one have you seen the baby seen the YouTube video of a if a video conference was in real life.
I’ll send you the like that dude it’ll you’ll die it’s you if it’s exactly the thing that you’re talking about okay what is your your your biggest personal Money Pit right now.
[34:02] Where you spending all your personal money.
Flavio Junqueira:
[34:06] I don’t know I think I’m just leaving that in the bank but I think I’m wine recently not a lot of it but I think probably more than I should.
Cory Minton:
[34:18] What region what region are you into these days.
Flavio Junqueira:
[34:20] Spanish wine of course.
Cory Minton:
[34:21] Mine are very cool I want you baby.
Fabian Hueske:
[34:23] Yeah I don’t have any expensive puppies.
Cory Minton:
[34:26] You guys are much smarter than us are you going anywhere exceptionally interesting soon for work or for.
[34:40] Shameless plug Fleet forward conference super interesting.
Fabian Hueske:
[34:42] I haven’t talked about that already and I’m not sure how exciting how exciting San Francisco it’s not a very exciting.
Cory Minton:
[34:51] Yeah there you go how about you you going to go anywhere cool.
Flavio Junqueira:
[34:54] Yes I’m going to Barcelona I leave there.
Cory Minton:
[34:56] You live in Barcelona I see I like I like Berlin and in Germany it’s a great sound Barcelona if I had to pick one city to live in in Europe I think it be Barcelona.
Yeah well so I got kind of thrown off because the the first customers visit that I went to an in Barcelona visit a couple years ago,
was Desi koala the the clothing manufacturer you know where their store is right over there their shop so it’s right there on the point and literally I’m in this conference room and it’s all glass.
I’m standing kind of on this whiteboard talking looking at people in the glasses behind them.
Animals on the beach and it’s the beat it’s the Barcelona Beach how do you guys get any work done or are you currently or have you recently binged on.
TV show or Netflix.
Flavio Junqueira:
[35:46] There is there is this show to series to see the series but that she is pretty serious I think when one of them is section on American Netflix is a cow money Heist.
[36:03] But I like that one a lot and does another one that I actually don’t know the translation to that to Spanish ministerio Del Tiempo.
Cory Minton:
[36:09] Okay alright do I need any TV shows are into these days.
Fabian Hueske:
[36:13] Well it’s it’s been awhile but I’ve watched like it’s going to be a crime show that was quite interesting I can’t remember the name of those.
Cory Minton:
[36:23] All right we’ll look it up what guys thanks again for being on we are so appreciative of your time here we hope your session goes well we wish you the best of luck will say break a leg and I certainly look forward to reading the book and I want to hear some updates out of 44 conference thanks again for being on John.
Flavio Junqueira:
[36:37] Thank you for hosting us.
Fabian Hueske:
[36:38] Thanks for asking us.