Merken

A Pythonic Approach to Continuous Delivery

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
thank thank you so we want to model and was planned on approach to continuous delivery and the I would fall beyond that we are company so of building data science applications and we also operate my job and there is some land and made them a lot of of the so called data services so we are offering some external data like web of public holidays and so on all those machine learning model to and answer yeah and as I said so and the main give a lot of and in addition I am also the main operations kind so the question is how to cooperate and double look at the same time and services with just a single person like me so more you can see that when we came up all of this could be done so all of you have heard of this it continuously varying just from the states maybe there's a strong bias if you if you so OK so just the of followed by the 1st of all I want I will explain what continues deliveries with some
definitions analogies so what my understanding is maybe it's not completely in agreement with every everybody else but my opinion on those of holidays such as degree pipeline look like and there we need the diversity is some boring the data then the biggest question OK I have been working Python code so policies not and there we read assemblers suddenly it's very the production lines of the delivery pipeline mode of some building blocks and as we'll we'll do in the High weight and then the output was possibly go wrong some tips and tricks from my side from my experience and maybe a short out what's might be the future some wishes and the short summary OK so let's start with what is
continuous delivery so I think to understand with continuous delivery is all about is to understand what it means workflow of traditional software development and here we have tools such so-called silos so we have a team of developers and we those cold start releases features continuous integration things like this in the end there's a product which is such a package with the version and on the other side of operations for operating points on that occasion and stuff like this so all the companies really doing and there we have terms like packaging and you always lifecycle configurations security monitoring all this stuff and so the lawless traditional way of looking at things all really hard to do with things was we have such a ball in between some some sometimes called the wall of confusion and those 2 silos and once the developers of finished with the new feature they just they get destroyed over the wall and say OK it's not problem now they should look how to deploy total cycle part of 1 of for our college configuring securities now what's happened is that
this is this new development of the province and this just means we see on this wall In this case all those things so it's it's just it's it's 1 thing so we have this we cannot be divided with the wall and say OK you do just and you that because it's just 1 thing we want to deliver this well so this is what 1 could call continuous delivery because now we have we don't have small anymore maybe with such another picture so I 1 could see continues serious as extending the development of cold conversions and stuff like this into production operations and also extending operations into the development of workflow and the important part is that now the development reading it would that the entire others stream and Value Stream is really just you have an idea you think our customers may like this 2 feature and then the value stream starts from that idea until it ends up in gives value to the customer and that's the important point is no wall in between just OK that's the value stream and development includes the whole values mean energy is important that we get the vector inside the stomach allotments like OK now to exercise
this continues to so all know know what you have to deliver value to the customer but continued delivery means that this is this child thing we want to release the people we want to really is often and we have this this thing continues is far more often than you think and this brings with it a real explosion of complexity Due to the increased demands on security safety they all called morning storing tests and this is the only possible by alternation so that any manual workflow just completely destroyed during continuous delivery continues yeah as as we know it is it is far far more often than you think you can do with just a little fox student for it's a new quality you really have to do it automatically you delivery of before you features OK this is there's another
thing is called the whole idea that of all value is any mechanism in the lean manufacturing process helped something and operator of all its purpose is to eliminate product defects by preventing correcting for drawing attention to the human eye terrorist testing of pure and that's may be such report point so we now need to leverage automation and automation something different and then the workflow of humans you can as humans can detect the also but just looking at things you feel there might be something wrong although it doesn't look correct and that's a completely different story once you go into the automation machines are found and that's so we really have to build in mechanisms to detect detect radius early on the various possible so
what could see all this stuff knowledge like such a company fracturing alternate Roberts production line and now you if you compare it to what we did so far is that traditional software development was just programming those properties but in the end we don't have any money if we we don't deliver past so that's just the new side of things this it's just a relevant program Roberts we have a good class of so now
let's look at the structural and
all out of like and then once you really do it you always feel kind of intersection jump and running so 1st of all need to change that is some software developers somewhere around the world committing something is deployed drugs would the exchange goes into production this no more on explain through someone who might be responsible for it and then decides on whatever all of those rule of I don't know so you just have to keep in mind each coming ends up in in production as soon as possible so we get this some time in between let's say 5 minutes later by the inflaton field committee will being semantic deposited and yeah exactly so unless it is it is proven to be not production ready and that's like in the structure and from running the school what we do is we design a challenges for such a change of subject matter and this is what you think and if we fail to detect overall change which is called you will end up in air and catch the princess so we don't want so we have to design challenges for this year yeah and there is more or less so I don't think that that it's possible this part of the bill and then have some other workflows manually as before so I would recommend to really start with the light with it a small what you automate the old problems so once you have such to walk instead of reading where you 1st just commit something and it ends up production so this is the 1st thing you should do and not to the solid start at the beginning and then we keep the rest as it was 10 20 40 years
so yeah you can see once again this jump and run the level that kind of thing so that so we don't call them developers anymore in contrast to operations just read the little value so it's the a of operations and development rejecting in maybe just the 1st thing that emerge conflict perspective that get back all them we have the traditional continuous integration which is really which proves the correctness of of the code maybe gathering and think that it is a 2 minutes later red of page fixing about committing again maybe just 1 goes through and then they are stages later on so the next thing would be an automated acceptance tests we will see a minute what is the what is in there maybe some acceptance tests of the weight so this is the same the minimal production line we read we have to implement and this picture is stolen from this book from just under so was reading very good it's school of practical tips and it's really really helpful
OK so there another picture from that book just to give you a feeling so that it's not that it hasn't to been that that's really all stages quantum state both trumpet around levels so and that they have to be streamlined they can also execute in parallel but the important thing is that naturally we want to increase our confidence in the change until it hits product so the beginning in the 1st stage we are not actually know maybe that all unit tests pass but this is really production-ready so this maybe still wore a forest green maybe that's not testify unit testing maybe it's important because it's all company and so on so going to the right means we have more confidence in the same it's to the right of the day environments where we test testing these things needs to be more production like so we have an environment which is called production this means median to instance on so it's a good idea to tried to the the best product as production like within the environment as possible in the early stages so it's too expensive located in to some of the trade of year and of course the gets faster the better than the old 1 someone committed 1st thing this stage this is the traditional continuous integration you commit and then the unit tests and so on and there we can be taken to mean that this increases mean the latest feedback that we get it from production and that's the most dangerous and most expect 1 OK so
now let's say you have some kind of working Python code all that application namely the world
what but so what do we do and yeah the 1st step is the need the problem development not to him and young so within transforming growth course we know let's call it the normal python package I think all of you know that there is no such thing as a normal part of the package they all different different versions tendencies and so on and so this is called the standard Python package but you could do anything else like proper Debian packages all some people stock of origin for the complete binary package of your artifacts or if you want you can do do it by hand was just some tough on this subject and some constraints should be unique version so that really is you know that's the number you should note that's the revision um should somehow managed to dependency so what do we expect our environment to be and the other is a small here so we can rebuild a small tool called price-controlled and there was a talk on Monday the last known packaging features and traits and was there they something similar as last slide so that was called the covered that later so this is nothing else than just the template of holiday to build standardized so at least for our company it's extremely important that all the pictures inside the company inside all production and delivery pipeline really look the same and the same means for example part of we see see so it's really easy you just say this is the most difficult to put up my hand and then you have a package which does not quite some things with some still want to make new versions of commit all the talk tests and springs documentation and you name it did repository is really have and there was just with these 2 lines we have 1st claim we have the problem so that the next
thing I think most of you know all continuous integration and yeah principle this means execute all unit tests does we do it in yeah so you can use any continuous integration tools for that report of drug use all there was it was all there this is of course it's really serve as a running somewhere so each comment is automatically is executed that's it is a very important that the reading if there is a common we really should know that latest 15 minutes later whether this committee was good anything else would you do values 1 thing of the young at least from my experience it's a good idea to already there start different stages reading it may be that some unit tests are running really really fast and others take let's say 1 minute it's it's already there best practice to a high so several jobs even get so that that you really have the fastest feet that you can get so after 15 seconds the 1st thing they condemned the long-running take up to 50 of OK so
that's yeah so as to have you if you're not good and if enough for those challenges all moment unit as we know all of them but it is and they are my definition would be the only test that you should go code there so we don't need any environment environment might be some running instances of services like this to read about it qualifies system maybe it's debatable or database stuff like that integration component state this how parts that can together and everything you databases maybe some small dummy data bases so that's up to you whatever you want and then of course something like statistical analysis might make sense highly accurate things like this I think it's a good idea to go over the edge of the coverage so that you can take there's the huge you of 20 thousand lines of code is 0 just 1 might also would doctors as a really good thing so that you just documentation that and what would we get all after seeing the continuous integration is is really a fixed object which means we over and we know this version of test with all continuous integration and for example in this scaffold have all 4 was 0 as a head full versioning sometimes the Python community decided that for continuous integration we really need for each commit another unique version it has to be somehow this way so we have the 1st 3 numbers are in the text so if you take the something and this is the there was this year after this fact because you never know the developing world and that's 15 means it's the 15 after the last and the thing under the is like good but the graph so it's quite easy to then reference from new version a when you find somewhere in Uganda this seem to want to what the Committee deserve of it's quite now it's now it's time to fill up all artifacts repository and maybe yeah OK maybe of book that you need an how to repository so we use apply for this yeah I think for those of you I don't so really data keynote this morning is that is the code available this so this seems to be no more than that 5 of it's that's what what would you know it's on training assault and so also solution for what this of hybrid or and has some quite good set of features so we we use quite heavily and making reference to the talk which different have also from young modes 12 30 1 and he really deep dive stick to their why so if you're interested it's really interesting what if there was something he sees a private this is an inheritance inheritance this is quite interesting
OK now that was the fun part
we all of so few of them the traditional because of beloved that's the part where you know what's happening is the continuous agree on and outcomes so what would I have used use and configuration maybe you can add packaging stuff but all of these dependent and 1 of the things you so
what all of for the next stage just think of this as the acceptance tests we really we need an automated knowledge because now we triggered genus integration and the this integration automatically triggers also acceptance then accept says we tested behavior of of all of of all service all went application or whatever so it's really we should designed and in a way you know you knowledge other users of the view that education and you really need to test how it behaves something like if like click on item and type my credit card thing here URI then I should get an e-mail saying that I bought this thing so it's really the forms that so everything you want your application to behave should be tested and for this theory we really it it has to be automatically deployed into some production like environment that's a possible in whole somewhere on the server of the median of what yeah and I think it's crucial to have read the same code which just automate deployed for this test it's the same to use later on the people into production thank is not a good idea to have this group which is called the acceptance test deploying and another squared which is more or less a copy this is called the drawing tool production because once you find the parking the deployed also the deployment is tested here so we have here is we are able to deploy all ratification into production so that it behaves exactly like what it's it's really hard to pin down that you also have to fix the problem is called people and so that's not what we want yeah the has that already in the environment and use those as possible and this is just my experience so you so somebody ask you what to estimate of how much time is needed for all donated deployed production and just doing multiplied by 3 no matter what you what you thought would you what you need for and so a I don't want to say here that should shouldn't doing it's just that it's not very small so pick up a really really small thing maybe where you think this should be done in 3 hours and then you need 1 so don't start with I think we should we should get this all running in 3 weeks because chances hybrid your management will maybe tell you after 3 months come on what's happening here maybe we have to stop and that would be so start small and then you get a feeling where all those things and what are the blocking things
it's worth reading this from advise you can you can do whatever we want which enormous shell script this is good enough why not it worked in the years before but configuration management was who knows what configuration management assessment tools are yeah just pieces so the users everything in bottles of magnitude but they are things like properties sold shares you name it and we use and so it's the pies 2 words it's written in Python you can think what it using extended as if you all not only high skills and I thonic all also because it's really really simple it's slightly and the very problem thing is decorative so you don't In in shell scripts you right to get that to that and he has more that's like history of you just say I want in the end you have this and it's really uses your going quite a lot and this
could be an example of a sensible playbook just to give you a feeling so what do I have to do to really alter my life going on OK we want to deploy something on the whole also called web service they are defined somewhere else some IP addresses the like that OK we want all of that to be installed we used foreign goes in what Durán and so we say the package name might have and we created with high all that it's already tested and it to the left and now we want to install it into a vector analysis so that we have not be messing up with some already installed systems dependency yeah and then we say OK as an index use all alone have higher because there we uploaded all package there is yeah it's it's maybe not a perfect example so really really once again if you don't just type it and use it for your production might collapse because doesn't mean this mean minus minus 3 this would also installed for each dependency development versions for example for requests and so on so maybe you should rethink and for example 1st install requirements and then just installing and your latest development and then you just start to at I I I did it also a decorative way because the shell script it's called my started if it would be a plane as their scriptures at the moment it would be called start my but then they get broken because it's running already so always think in
declarative description OK so acceptance tests as our said already carried proved to to be able to really be careful you that's your money it's not it's your last chance it's really defining was of Office this thing and there are tools they are really some some tools was like the rate for example they are building except for these kind of things is by who denied in the technology and press buttons text and you write it really that sentences so you know management can write those acceptance tests and you execute these like 1st see in your voice we think like this but to whatever you want to
and that's it so that's the that's the last step so you have all those acceptance tests of past that's the last from you might want to have
some additional nonfunctional things these are things like formants measurements so that you don't have any surprised to see that the check out not takes we already have security and maybe explorer that might the no the feasible with manual test so really if you if you have an experience test the only tries to screw up the of you think but that that of course is you have to wait until the test this time to test so that might if you really have a you really have to wait for it that would be such a Brookings thing onto the test stimuli that at that time and to do this take but fictional you will you read on to have such a manual approved so that would be such a popular where someone is pushing the button but it is not fall of taking over responsibility to some managers will then it's fine if there is no that's but it's more or less for things like the 1 who works in the marketing campaign because you've each it's really really breaking change so you want to coordinate somehow made of some other things legal issues maybe other allowed to some information early on and if possible you can do things like there are many different possibilities can already releases that's of something that Facebook does that maybe you can already give some beta versions to some some this maybe randomized fault you select from users can already do it so that you can do whatever you want what we do is we just have want continent and OK deployed how
the important thing is the but if it gets quite complicated to of all all those stages and this workflow which is completely optimized you don't have someone who says I we you we should not forget and until next Friday to execute X 1 so you need some all coordination past unit tests passed integration press passed acceptance tests security testing things like this that and now we we do it with with Jenkins there are some specialized continuous delivery to which I know from all works it's goal don't see you all this idea and urban code I never use them because we knew we we just have Jenkins already there so we do those jobs together and this is used as a tool called what's deployed problem the good laughter as 1 as the but what populated assessment so there seems to be an issue with the latest tuple suppose that so it should be covered some of our early on so these tests was that he yeah so all manual approval here is to click on the change in the bottom and now yeah this means deployed to production so you see already it's not really specialized tool for that so build and deploy to production maybe something but it
works so question is what could possibly go wrong but this is it sounds quite so much of
the illegitimate but it's not so all my rights as we keep everything every every everything simple stupid don't do any complex things you will screw up so we always if you have if we had an issue somewhere it was always the 1st what version was too complex and it is a real dummy implementations this indicates that the simplest solution and this is always the you have to
automate all the things read all OK I do it because I'm really lazy and doing the things twice this already for me not acceptable but there quite some other arguments for you complete delivery pipeline from idea until the last part is simple version control so you have for example predictable recovery so you know the 1 minute OK so you know and and the sun goes down you can deployed somewhere else in another region because you know everything is in version control does nothing and which is about trust that in such restricted to the past like I have to we could figure out that scripts in order to be a machine suggest that they don't do with arrows and repetitive tasks and you can concentrate on the delivery so you can think of just which might bring value to to the cops was that the whole point that good to reduce the maintained rechecked of real time for you delivery pipelines for you automate deployed you have to migrate to new versions off and so you have likely to move versions of requests and so on so really do it don't don't just think of the red once and now we can stop automated never is automated we have continuously improved the deployment and are so the class things that really anything so that if you have to do is you have to work I think it's it's quite high art what mission if you've already knew that feedback that would need to have the right to get that's it's really not a good idea so but later have some Cloud thing there where machines can automatically get all the things automated with DPI things like so the future now
OK ready OK so we've learned from from you know he's the same their opinion and mean dependency management packaging and all this stuff is somehow not so perfect implies so we we have to to find ways to make it better on if if you really you automatically deployed to production after each Committee you really want to know what's installed there are positive is now a new version this morning and will the deployed once again and it's almost a summary of a new version over there so it's it's not really it's not really perfect this the 2 worlds problem we we have for example the general act is installing pop off a few dependencies but that constant and installs another half maybe there were 2 and was completely different versions of all the dependencies that's not not perfect and and this such python let's say lightweight and easy to use uh continuous delivery who was also missing 2 which is really away after delivery pipeline which knows OK this is the acceptance test stage and we deployed version number x in yesterday's by he was among those that and that so that's still missing and many many tools we may need to was optimized for menu workflow which means there is still operations guys sitting on the terminal typing in that installed so over Jenkins type wisely good immediately immediately broke that across have to effort to which types in yes for you In the nonexistent terminal and that's that's that's not because so we have to whether might we have to act on that we really need to be improved to so all
results are real these media is the dropped since it's edge i with feedback from the customer so we don't lose money automated is that the amendment of collaboration that in some of those you can do it just stop export of building block once again and ask for competitors that of capital repository Jenkins continuous integration hearing of the 5 lines of Python unit tests that's at the will stupid tests and hence automated lost and you really need to encourage you want to do it and commit something and it ends up introduction to so look
something new what did you
few few but will want day and 1 would question the 1 shot question have you heard of the Jenkins was full of plug-in I heard of it but I have never use use because then Jenkins really the continent this delivery system not only the status of striatal up with with having singled stops that are connected think you're connecting the cost manually more or less flat on the workflow plug-ins really again change of course the irony I would I have to look into it's a difficult thank you of because in this few if you have any questions we already beyond the boot low and over there
Nichtlinearer Operator
Virtuelle Maschine
Addition
Benutzerbeteiligung
Informationsmodellierung
Web Services
Prozess <Informatik>
Kartesische Koordinaten
Computeranimation
Aggregatzustand
Gewicht <Mathematik>
Total <Mathematik>
Punkt
Desintegration <Mathematik>
Regulärer Ausdruck
Ikosaeder
Term
Code
Computeranimation
Softwaretest
Code
Gruppoid
Softwareentwickler
Konfigurationsraum
Analogieschluss
Gerade
Funktion <Mathematik>
Nichtlinearer Operator
ATM
Softwareentwickler
Assembler
Computersicherheit
Division
Kontinuierliche Integration
p-Block
Biprodukt
Gerade
Minimalgrad
Mereologie
Dreiecksfreier Graph
Versionsverwaltung
Binärdaten
Softwaretest
Nichtlinearer Operator
Umsetzung <Informatik>
Punkt
Desintegration <Mathematik>
Computersicherheit
t-Test
Vektorraum
Biprodukt
Komplex <Algebra>
Computeranimation
Streaming <Kommunikationstechnik>
Energiedichte
Trigonometrische Funktion
Softwaretest
Rückkopplung
Reelle Zahl
Code
Mereologie
Softwareentwickler
Streaming <Kommunikationstechnik>
Versionsverwaltung
Ganze Funktion
Softwaretest
Nichtlinearer Operator
Radius
Kraftfahrzeugmechatroniker
Subtraktion
Punkt
Prozess <Physik>
Sechsecknetz
POKE
Klasse <Mathematik>
Softwareentwicklung
Eliminationsverfahren
Nichtlinearer Operator
Biprodukt
Speicherbereichsnetzwerk
Computeranimation
Äußere Algebra eines Moduls
Biprodukt
Softwareentwickler
Gerade
Verkehrsinformation
Binärdaten
Wechselsprung
Datenfeld
Skeleton <Programmierung>
REST <Informatik>
Mathematisierung
Mereologie
Mathematisierung
Schlussregel
Softwareentwickler
Biprodukt
Computeranimation
Rückkopplung
Gewicht <Mathematik>
Komponententest
Mathematisierung
Versionsverwaltung
Code
Computeranimation
Homepage
Übergang
Metropolitan area network
Einheit <Mathematik>
Bereichsschätzung
Perspektive
Kontrast <Statistik>
Softwareentwickler
Gerade
Softwaretest
Nichtlinearer Operator
Extremwert
Wald <Graphentheorie>
Güte der Anpassung
Kontinuierliche Integration
Biprodukt
Medianwert
Teilmenge
Rechter Winkel
Programmierumgebung
Aggregatzustand
Instantiierung
Quelle <Physik>
Softwaretest
Nebenbedingung
Subtraktion
Dokumentenserver
Template
Versionsverwaltung
Zahlenbereich
Ähnlichkeitsgeometrie
Kartesische Koordinaten
Biprodukt
Code
Computeranimation
Rechenschieber
Arithmetisches Mittel
Code
Eigentliche Abbildung
Mereologie
Softwareentwickler
Normalvektor
Programmierumgebung
Gerade
Standardabweichung
Offene Menge
Komponententest
Wellenpaket
Momentenproblem
Dokumentenserver
Desintegration <Mathematik>
Versionsverwaltung
Zahlenbereich
Code
Computeranimation
Physikalisches System
Web Services
Einheit <Mathematik>
Prozess <Informatik>
Vererbungshierarchie
Zusammenhängender Graph
Ideal <Mathematik>
Gerade
Schreib-Lese-Kopf
Softwaretest
ATM
Dokumentenserver
Graph
Datenhaltung
Güte der Anpassung
Zwei
Kontinuierliche Integration
Statistische Analyse
Physikalisches System
Integral
Objekt <Kategorie>
Arithmetisches Mittel
Menge
Mereologie
Ruhmasse
Datenfluss
Versionsverwaltung
Programmierumgebung
Verkehrsinformation
Instantiierung
Aggregatzustand
Lesen <Datenverarbeitung>
Softwaretest
Sichtenkonzept
Gruppenkeim
Kartesische Koordinaten
Programmierumgebung
Biprodukt
Medianwert
Physikalische Theorie
Code
Computeranimation
Integral
Chipkarte
Teilmenge
Bildschirmmaske
Softwaretest
Datenmanagement
Web Services
Mereologie
Datentyp
Server
Konfigurationsraum
Programmierumgebung
Ebene
Größenordnung
Nabel <Mathematik>
Momentenproblem
Gemeinsamer Speicher
Versionsverwaltung
Datenmanagement
Computeranimation
Vektoranalysis
Task
Web Services
Datentyp
Statistische Analyse
Pi <Zahl>
Skript <Programm>
Softwareentwickler
Ordnung <Mathematik>
Binärdaten
Videospiel
Kategorie <Mathematik>
Physikalisches System
Biprodukt
Nabel <Mathematik>
Arithmetisches Mittel
Portscanner
Konfigurationsverwaltung
Automatische Indexierung
Wort <Informatik>
Größenordnung
Portscanner
Softwaretest
Teilmenge
Deskriptive Statistik
Softwaretest
Datenmanagement
Fächer <Mathematik>
Bitrate
Computeranimation
Office-Paket
Softwaretest
Facebook
Prozess <Informatik>
Computersicherheit
Mathematisierung
n-Tupel
Versionsverwaltung
Biprodukt
Code
Computeranimation
Integral
Teilmenge
Metropolitan area network
Einheit <Mathematik>
Datenmanagement
Prozess <Informatik>
Code
Minimum
Endogene Variable
Information
Versionsverwaltung
Personal Area Network
Koordinaten
Einflussgröße
Portscanner
Rechter Winkel
Versionsverwaltung
Implementierung
Knoten <Statik>
Computeranimation
Wiederherstellung <Informatik>
Webforum
Rückkopplung
Punkt
Klasse <Mathematik>
Versionsverwaltung
Datenmanagement
Zahlenbereich
Computeranimation
Wiederherstellung <Informatik>
Task
Metropolitan area network
Virtuelle Maschine
Prognoseverfahren
Datenmanagement
Momentenproblem
Datentyp
Radikal <Mathematik>
Zeitrichtung
Skript <Programm>
Installation <Informatik>
Softwaretest
Parametersystem
Nichtlinearer Operator
Vervollständigung <Mathematik>
Knoten <Statik>
Biprodukt
Portscanner
Teilmenge
Arithmetisches Mittel
Echtzeitsystem
Rechter Winkel
Mereologie
Wiederherstellung <Informatik>
Ordnung <Mathematik>
Softwaretest
Resultante
Rückkopplung
Dokumentenserver
Dokumentenserver
Kontinuierliche Integration
Computeranimation
Portscanner
Trigonometrische Funktion
Kollaboration <Informatik>
Einheit <Mathematik>
Softwaretest
Iteration
Rückkopplung
Hypermedia
Gerade
SIMA-Dialogverfahren
Systemprogrammierung
Mathematisierung
Plug in
Physikalisches System
Computeranimation
Haar-Integral

Metadaten

Formale Metadaten

Titel A Pythonic Approach to Continuous Delivery
Serientitel EuroPython 2015
Teil 132
Anzahl der Teile 173
Autor Neubauer, Sebastian
Lizenz CC-Namensnennung - keine kommerzielle Nutzung - Weitergabe unter gleichen Bedingungen 3.0 Unported:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen und nicht-kommerziellen Zweck nutzen, verändern und in unveränderter oder veränderter Form vervielfältigen, verbreiten und öffentlich zugänglich machen, sofern Sie den Namen des Autors/Rechteinhabers in der von ihm festgelegten Weise nennen und das Werk bzw. diesen Inhalt auch in veränderter Form nur unter den Bedingungen dieser Lizenz weitergeben
DOI 10.5446/20201
Herausgeber EuroPython
Erscheinungsjahr 2015
Sprache Englisch
Produktionsort Bilbao, Euskadi, Spain

Technische Metadaten

Dauer 39:11

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Sebastian Neubauer - A Pythonic Approach to Continuous Delivery Software development is all about writing code that delivers additional value to a customer. Following the agile and lean approach this value created by code changes should be continuously delivered as fast, as early and as often as possible without any compromise on the quality. Remarkably, there is a huge gap between the development of the application code and the reliable and scalable operation of the application. As an example, most of the tutorials about web development with Flask or Django end by starting a local “dummy” server, missing out all the steps needed for production ready operation of the web service. Furthermore, as there is no “rocket science” in-between, many proposals to bridge that gap from both sides, operations and developers start with sentences like: “you just have to...”, a clear indication that it will cause problems later on and also a symptom of a cultural gap between developers and operations staff. In this talk I will go through the complete delivery pipeline from application development to the industrial grade operation, clearly biased towards the “DevOps” mindset. Instead of presenting a sophisticated enterprise solution, I will outline the necessary building blocks for continuous delivery and fill them up with simple but working poor man's solutions, so that it is equally useful for professional and non-professional developers and operations engineers. After the talk you will know how to build such a continuous delivery pipeline with open-source tools like “Ansible”, “Devpi” and “Jenkins” and I will share some of my day-to-day experiences with automation in general. Although many of the concepts are language agnostic I will focus on the ins and outs in a python universe and outline the pythonic way of “get this thing running”.
Schlagwörter EuroPython Conference
EP 2015
EuroPython 2015

Ähnliche Filme

Loading...