Bestand wählen
Merken

Introducing Delivery in Enterprises - Lessons, Techniques, Tips and Tricks

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
so yeah I'm this is me obviously but I am very see stars at their what's practitioner their votes evangelist um I I'm actually 8 squat ops evangelist 2 slightly different but kind of the same thing systems quality users and that what's on and we talk about us and what I did with my previous company and no longer bear so if I don't mention names I apologize that's just always it's going to be the this is a title talk Reshef delivery on enterprises tips tricks and techniques this is going to be a little bit more on cultural than it is technical of the space Outline which is you and talk about the base state of what you know me find in an enterprise I will give a talk about how you get buy-in from people how you do a proof of concept the workflows and yes mitigate tation is specific to spell that way because we have to worry about it on then how we fix things in our chef environment and cottage of environment and also some Konishi CC with delivery and then this still refer to it as delivery even is now suddenly become workflow on thanks the completes screwing up all my slides today I of his a really basic if you don't wanna be here you I should just killing time and this is what you're going to get to hear about right on basically an organization is made up of people the technical issues are important they're summary serious ones we touch on the beginning people on board and more importantly educated is going to save you time money heartache on an effect of a lot of pain sister right on and on the base state yes everything's kind a comic font thing so he's going to deal with that today on qualities is a apart most people in enterprises think they have some kind of continuous integration who here has like continues inversion actually running in the enterprise a TTS how many people actually think kicking the hands how we could but something that's more continuous integration H it's more like you know OK and how many people for cancer then continues integration means well we got Jenkins running and the bill when red that's good and bad Sicily at 2 or 3 people that OK so here the normal continuous integration problems right we have a large number of the tests that are running contains and somebody's a development test some of these are just tested really work and never have worked on this bunch of information that comes out from the from the the continuous integration box and tells you a lot of stuff none of which you really need to know it affects your arm picture Tesla fix what's going on that noise causes continues integration blindness but 1 thing you have to overcome when you're going on the route to delivery the scale of your test infrastructure the bigger the test infrastructure gets them unless you get the feedback because you get more more noise right and unless you're actually sitting down going OK this master will talk about solutions this massive belongs to this team you can have like really long feedback to really why committed to it I don't have a very large test infrastructure so we know it's going to take an hour for runs all come back to that martensitic if it passed or got like a really really large sets infrastructure be but everything we know the other thing happened and this is fairly true in most enterprises of the code goes over the wall to QA doesn't really get tested code is not tested by developers order is thought of as the by developers but then there was some my box perfectly fine right so here's how we solve those right 1st thing was to clean up the notifications I we took down every single e-mail and made it into basically 4 lines what went wrong where you go and find the consul output the why you were getting this what machine was around we clearly defined at test infrastructure every single team got a master right and that really helps means every silting has its own mass it means that you are only listening to those machine we didn't get to do this is is it contains all the way down case we did not get to do something that I'm I'm hoping to implement of my car which is that every single 5 team gets a Jenkins master in a container which means that every single developed can use that Jenkins master to run the tests locally and then normal Jenkins latent in of container as well this goes the same for Travis is goes the same for any of the sea ice and that way you know that what happens on your local box is going to happen when it hits a CI system and then we gain if I'd testing as well we had like a nice little thing whereby you could see which team was to invest in the number of tests that was running or how long the specific without continuous integration you don't have continuous delivery without continuous delivery and you you know it's British delivery here without consist of you can't move forwards right you can go down the path a ship delivery if you don't at least have some basic things when we talk about this OK this is fairly true that the
power resides women believe it provides no more no less that means that it doesn't matter how much you want delivery does matter how much you wanna get on the ship automate rock unless the people such as signing the check signed the check you not going to get it right so we in this room are primarily in the middle circle here right where the implementers of change where the Russian be doing the work but we don't work very well unless we have some guinea pigs to run a changes on and you can treat them like the think of like the the nicely and make sure the most offending so much of and then there's the wielders of power right those the people who really when it comes down to the by the got some political ambition of the company for they've got some actual real money all they want to be seen to be like leaving change in making cultural changes with the company of this that's the 1 example whereby a person literally left the company for a period of 3 weeks on the same molecule get delivery and and got everything up to the door right could not get it I knew that aligned compared the line they went on vacation for 3 weeks they came back and miraculously the person who had been going there we should have we should have you should have it was like I have brought delivery to the company that stuff does happen it's kind of painful and is part of the culture that we need to count start to break down and 1 of the ways we break this down it's called bringing in the holding in the helps red brings in collaboration tool so spoke to at last count 14 different companies have been working with delivery this thing here squad ops is the thing that makes things worse when they were bringing in delivery they made it seem that was focused around not only the systems but the quality of the users and that what states it wasn't just all the DevOps thing is going to go do this right a fully cross-functional teams OK is creating culture and that's what we're doing with delivery it's data-driven as much as anything else right they want to be able to show pain this is exactly how this works here the things we're going to show and this is how much risk you're going to find get so plan out the benefits of how you're going to show these things right we know that we get faster reaction times continues delivery right we know that all it doesn't matter I'm going to do this thing for like it's nice to have featured the were about it will be a sum becomes is how we must have that feature right now right or you have a new customer comes on board says love everything you that they've done so far I need everything to be multitenant right now right and suddenly you got the that we consider the the have the ability to just can I go I mean shift liver-specific they have the ability to go OK we can adopt we we can adapt to that so you prove concept needs to show that they also needs to show how little risk there is when you during a Release and Deployment so we have 80 to 100 cluster of fleet that we would deploy to every roughly 3 weeks that cluster was about 6 to 8 hours to roll out this is horrible the reason is 6 to 8 hours was because somebody had not actually decided to put the genomes on the boxes just pull down every single jump every time I had never actually not some of the gems and so Ruby had to be restarted every single box every single roll out every 3 weeks Kenny said take 6 hours if you're doing that on a daily basis or multiple times per day that can happen right so you fix that up front and of course that's an efficiency that you get to expose if you can show this to your and the powers of the power wielded and that's what you actually end up doing in going look here's our risk here's on efficiency we can solve that and this is the way in which just live with other the other thing about chef
deliveries was allows it have much more flexible releases thank you or your source code and under test code and you automation all beside together so all of integration is that right and because the system is there to be continuously delivered right it means that the system has to be flexible to handle anything that's coming into it even those 1 might solid pipeline right it still needs to be flexible enough to be able to manage everything going out so this is a great things to show and and move on to this
right hey let's go for it delivery OK so now you need to build out the proof of concept right that was all my planning and they expect to start with OK 1st and think about when you think about building up a proof of concept is environments environments are not as much as we want them to be they have change they are no longer the it's this set of machine all this set emissions right it's a set of services that talk to 1 another right that set of services talking to 1 of those effectively configuration that if you start planning your environments that way of thinking all you know doesn't matter that resides in this data center and have that resides in that data center i need these things to toll properly because that's actually my acceptance environment for this service and those the the dependence that that's the way you start to think through you already have a set of this working life in every single enterprise of ever spoken to has already we have staging you may not call a staging may decide call test to on you may have production which is know there's a cold beta so all their various other different names but a set of environments now as a question but figuring out which of the environment services you map right and that really boils down to a really huge problem with delivery not a problem at delivery itself but with adopting and that is the network topology delivery has to be able to talk to everything is going to manage right and it's going to manage for environments up front and manage acceptance union rehearsal and delivered and here's the really cool thing right even if you have got next staging staging beta staging in production start talking about these names the from day 1 no OK acceptance 1 is acceptance acceptance is basically isolation testing right union is basically a big integration of the pipeline rehearsal it's the pilot stage and then delivered his production or wherever you want to go with the production of we found that in on cases where people have actually done this in USA delivery turns up front and much more successful adoption rates now crank through this I think everybody here knows the the benefits of using separate BPCs 1st separate environments if you don't his can have a conversation with the upwards of security groups this is kind of important it's difficult to move security groups with Amazon classic obviously on it's kind of difficult thing to do and you will want to do that what you've got your preferred for concept out so trying to figure out how you can do that for real and again magicker environments well thank we talked about this a little
bit at least 3 of the companies I spoke to you have to build bridges between different environments in order to get a delivery to work spending the time up front to work on the network topology of the graph paper rather than have it all residing in E C 2 and I've got some of that machine so maybe after like figure out some security groups and imports here in Figure all of that stuff out up front it makes things so much faster so much more but I see sect there and that's going to turn around so it doesn't like this piece on but co-locating acceptance and union is actually useful a lot of times you have depended Services Independent Services acceptance this specifically supposed to be more isolation testing right you got service a depending on service be if you have service a continually ramping and service the continue ramping which one's the but line you can tell especially that 2 separate repose 2 separate into developing on that means you have a really hard time getting down rest the pipeline and so everybody that I've spoken to right you got this working really really well had at least some kind of data store they had an acceptance within a cop some data from union back into the data store so they could actually use it for tests right again we haven't even got to the point where actually building this out we just know that these are things we are that we need to have a requirement this is a proof of concept spot real the I know he wants to be realized by and yes it is the real thing that people here have lychee got it's a prototype or just read out to so it is good and then found out like 2 months 3 months a year later that's actually still out in production and running right everybody does it figure out the a proof of concept is basically your 1st MDP right it's going to be the real thing you may expand that that is going to be there for as long as the liver is around you can so the proof of concept itself I forget who wrote this quote on the best way to predict the future is to invent it so that means that we do that right we go OK it's not just a proof of concept where actually make sure that this is real you can show that make inflation rounds well some choices like this liver the embodiment of culture that the the liver the delivery allows you to empower teams and collaborate really really close down that central pipeline right because that's the case when your culture is going to be scrutinized during the during a proof of concept it doesn't matter how on board your tenancy there's always in so go well it's got to really work right the semantic side delivery is the best thing since sliced bread for know team a team these going you have this it really work in my case how they need to see that and indeed see the culture is going to to benefit as well right the 3 things that we really talk about there's communicating and collaborating automated that means that you communicate every single time point going forward because of all done this is how we doing this is the right the collaboration pieces many people in his unique even if it's just like half a day but it's amazing how many people of you have been to a dragon for half a day to help set up the repo all set up a bit bucket integration or deal with like the owner of particular project and then gone back to the teens and on hey this than everything is actually pretty cool usually is set up and then you can automate that stuff especially when you view showing it to someone it's way easier to automate something that you explain to somebody else that is to automate something that you kind of doing your own way you said there you go I think it works this way I think it works this way this automated always that well this is no manual piece of the beginning and you have to hit complaint about in the middle and at the end this kind of got there and have to do like some random or sink Hank it's much easier when you explain it to somebody to go all I really want only that these that these that these we talked about assumed before creating cross-functional teams and to build that that is really really important so that cross-functional teams really should look like this operations test told deployment and a key engineering team it is so so so tempting terrorist all you know what that this little project that doesn't really do anything let's put that on delivery and we can use that as I think the guess what they don't do very much so there's no real gain and there's no real benefit there that we find a key team that is an especially large size that actually has cycles of iterations you can go through the but it's small enough that you know all the problems of having to deal with 57 different project managers 104 indifferent about everything else kind of goes away and it's a long process guides sorry on the average proof of concept take from 2 months to install and get running usually because people don't have teams ago of this is what you do and do you build it over to its effect on gallate of John Morrow drew this John on they helped us out a lot on my last place and we were up and running in 2 weeks we can say yes the performance of the 2 week forgetting other teams to use it and showing the data flowing through that was about 2 then there's what i'd laughing you referred to as the gestation period and the gestation period is effectively people go on at school that we should that lead to more about that and then nothing right i and that the impetus of the proof of concept to the period of going yes we're going to really build it is around about my months on right the full adoption across the value of so you're looking at somewhere between a year a year and a half to go from by delivery to we're using delivery right is a long time but it's the time we are educating people under showing exactly the benefits it's also time where most of you will be using the kind we start work early on these things we demo demo them over time then you actually short no cycles we found that most people who have done like 2 weekly strength and that a proper demo at the end of those 2 weeks had much greater velocity to the project noted they have great of to watch it they kept the velocity after the project was built it took that exact same people back to their normal normal to the right and just dealt with that it's like all this is how we do this here this is much better and men the collaboration across the across the organizations with that we know this Strider delivery is the single pipeline right of this little thing job here is the deployment of how many people know and endangered employers right but so we just dropped some box goes whom I'm just gonna put this piece of code then now because that works OK cool and doesn't tell anybody doesn't like Jerry ticket doesn't writer uh deploy script doesn't do anything right let's get rid of man to deploy user pipeline everything goes through that that means you cookbooks go through right that means your applications go through means any test changes go through all the configuration and and because the team is likely to be building up the proof of concept for the delivery and place is the operations in the 1 I just musical works right up front right that's exactly what we did on back the previous placed on we had I think 213 cookbooks moved in a day and a half In this 230 because of of of thank we added extra benefit of doing that
is you all asking people to change the you're showing people thought what the change is and how it's beneficial right you not telling people you now have to do this you say hey look we've done this and it works really well right that way everybody gets to see it all which brings us to workflows and mitigate tation OK if you're not and FIL if you just operate by having know really it's more if you just deploying the way you to that you're not really thinking about how the point of this work right if you're just working along well I just do it commit minus a and close git push right you're not thinking about exactly what's going on right we are asking people to change their workflows with delivery lots of people come from get float to on delivery a lot of people don't from straight get to liver and every single 1 of them has the same thing what happens was interesting get pushed and then and delivery in it and and delivery review and what where where what is this right so we found the people you talked about delivery as a system for allowing code to flow through the development from development to production were much more successful than the source control is about controlling the history of your source and figuring out what the change was the liver is about getting back changed from a developer to a user right if you think about it that when you explain it that way people get on board way quicker right you already have workflows infrastructure that already exist so you have to map to that and kind of make these things when the changes are really visible so people understand the
biggest change is using this using delivery as the code told there's a reason that we have to get out integration and there's a reason we have a bucket integration and that's because most enterprises already have or bucket or some other system right and they used to using those tools as code review but I can't say this is not rely on the tool sets that you've got that you know work make them better by using the word that's the way to approach the problem the way that the people approached it successfully right it means you don't get swamped by all knowledge but I really like doing might create pull request and in BitBucket and I know exactly where but my but all my bookmark but not set up and and all that stuff and not go to this other place no you don't you to do the same 1 we talk about this a little bit as well and can kind repeat some things as I go through this make sure that the delivery is control its control of the source flow of the code flow from the developed to production delivery allows you to actually control that in a way that makes sense for the business unfortunately or fortunately do and how you look at the business accounts right source control as we call it in terms of like get is not just knowing a history on it means that we actually have to control emerges we have the and figure out exactly what's going on the right this random commit the do equal death so modern masters of delivery converge to master that is like the biggest thing that we got from every single person right get enterprise do exact same thing that but we do exactly the same thing at the crank a little bit further faster education saves lives depends again we talked about this year talk about the points where delivery exists as a system and it exists as you become right have workshops workshops were really really well actual practical hands-on workshops right for all of the teams once you go that that route concept you get you get people in there OK so let's move 1 of your because across right into a natural hands on workshop with them and show them where you get push doesn't work and you want to use that we review the biggest thing we found that we recommended to people on and this has been recommended on the on the ship success channels as well as externally right is you can create get extensions to really easy to create right the rat liver if you really can't get people to buy into it and your whole system is going that way right it's worth spending the 2 or 3 days to write write those will get extensions and OK this is how you do it right it stops people who are new to get getting overwhelmed and stops people who have been dealing with it for so long we just want to be with get like don't plan I don't want to do a delivery review again this is
not the type of delivery truck is a feeling that is not your friend is your friend if you're going straight into delivery and you are completely brand new and you have your own brand new chef environment right that every single enterprise will end up hacking delivery truck to some extent if you're talking to multiple different masters you will probably end up having to have delivery truck the every once know exactly how and in a separate my environments right checkout already done deals with this right it's environment right then have delivery truck go back through the pipeline again once you've proven everything work just push that 1 change up the environment right and in the real world again this prevented the into deployed 2 canary right most big enterprises using delivery are not going fully out of production the going up to a small subset it in by the data this course keep keepership versions law right of the so that everything I research server is the same right you don't want to be doing what we were doing which was an chapter 14 for delivery in chapter 12 the rest of the infrastructure causes all sorts of weird headaches that you would expect and so upgrade your kerchief infrastructure if you can then I will use multiple deployment methods from every company again I spoke to has at least 2 or 3 different flavor that some people just using CAPS scripts some people are pure are sinking I've seen people doing horrible deploys all of right so if you can this is the habitat piece right if you can try and get the the um the trying it habitat can built up front the habitat package builds up right front and just pass that down to the delivery pipeline right makes things a lot easier if you can't deployments a part of the source so it doesn't matter you can have every team doing a different point that providing the use in building provision the same way the the and yes it so that the company x on Jenkins was not used just the CI we use that also to do our uncontainable manage the security monster up from right to have a conversation upfront you will always always always right and that's why I don't really want to be continuously delivering to production you continuously delivery man but that's the kind of mentality that people think about so it's OK I can get out to the pre-production it out to a canary right and this is actually fairly normal model it's not a universal model but it's very it's very widely used to the issues the log and the
delivery section of the environment file you need to have the fully qualified domain name and that's not just your IP address important that you find that you like installed everything and you still can't get a login script that's likely to be of right on if you need to migrate a project from delivering the built in a proof of concept to bucket it's actually a lot easier than it looks when you read the documentation you can just update the project setting and it works and most of the time if it doesn't there's probably just something slightly tweaking in in the right my no don't got that again the Jeff nodes need to have a gene's on them right so you can now update the metadata to actually bring the gene's over into Chapter and effectively so you don't have to keep pulling down at this time because of the time you pull them down it makes things a lot longer and we talk about this continues the rate of production is scary 1st set delivered to a canary cluster member delivery truck a common issue that we had across the teams was that you people sort of run felt they didn't know why right a lot of that logging onto which a lot of the node and checking exactly why that's that's there right we found that a lot of the time you shift provisioning rather than the provisions that of delivery you could easily see what the error what was happening at that point this 1 waiting on worker I think we have that for thank 9 weeks in a row where we just have something what that's OK but if you can't think if you're confident that Jason is is incorrect and balked you get this is better has to be to be fair right there when we when we 1st got it and the reason that the 2nd I love about so we were the 1st paying customers for the deliberate when we 1st adopted it just that they're just went on doing something you have no idea what the least now you have like some kind of vague I'm waiting for work of but all the workers were free so you can go to it's very difficult to do upgrades of delivery if you're behind a corporate proxy there's been a lot of work recently on prox isn't as on certificates right it really helps you have to dig into that documentation upfront right if you get your self aware of how that's going to work you will be much more likely to adopt easily right you begin to explain those things to the security presence the wife got the school proposing nothing can come in needs to upgrade but nothing can come in and you need to have that conversation of the OK and those are the same people that the talk about security controls right and they say that you can have richer we ended up writing and at least 2 cookbooks that just granted sued to access to the delivery and the said they you know you do access and for every environment except for production on because we were going up production but this is a common complaint that we've seen some how pipelines can have at their house enterprise it's 1 per project that's not the case in the buckets are admitted that but not so for computing on but be aware that upfront you know have multiple pipelines per project with with the Gavin price thank you that's at the big link up here actually the crib most of my stuff from that right this stuff actually works the documentation for delivery is insanely good and it's readable by people like me as a manager as opposed to people who are actually doing the work right everybody gets deleted call thank you very much a few this time for questions if you want on and I'll be right here in yes Meyer and that kind the the the and that what so what i k so this is this is where I jump up and down and say that there is no such thing as their thoughts and DevOps is dead of we we have unfortunately build silos out the thing that we're trying to get away from with their votes is the silencing of a business you try get more more collaboration right unfortunate name Dev Ops means it's like always development operations right I there's a number of posts on this uh not least of which is and if you have a debt of the units daft idea you should feel bad I devils is a culture of collaboration across your organization if you want practicing shaft evolves kung fu you not talking about things like sales being involved in finance being involved the project management being involved right they're you really not practicing true Dev Ops you got a definition that what piece of it and sources sound preachy this is just call it my little and high costs there's a reason I use QuartOPs most of the time rather devil because that way you actually have people buying into the system the quality the users and the above operations people right that means everything every single person seemed to that means your cross-functional teams half to be involving other parties you can just go I've got this development operations team of its engineering team and do that together because you don't get the visibility so I I I completely understand it and the devil portmanteau we have to start thinking about how we express that white the more training more the malignant it from you in this that that that when he few
Bit
Abstimmung <Frequenz>
Aggregatzustand
Raum-Zeit
Kontinuierliche Integration
Eins
Stetige Abbildung
Softwaretest
Font
Maßstab
Code
Umkehrung <Mathematik>
Analytische Fortsetzung
E-Mail
Gerade
Funktion <Mathematik>
Softwaretest
Umwandlungsenthalpie
Nichtlinearer Operator
Zentrische Streckung
Ruhmasse
Programmierumgebung
Rechenschieber
Menge
Rechter Winkel
Beweistheorie
Information
Ordnung <Mathematik>
Programmierumgebung
Aggregatzustand
Rückkopplung
Quader
Selbst organisierendes System
Zahlenbereich
Geräusch
Code
Unternehmensarchitektur
Virtuelle Maschine
Geräusch
Selbst organisierendes System
Softwareentwickler
Soundverarbeitung
Kontinuierliche Integration
Routing
Physikalisches System
Integral
Whiteboard
Rückkopplung
Normalvektor
Unternehmensarchitektur
Stereometrie
Gewichtete Summe
Quader
Mathematisierung
Implementierung
Zählen
Code
Whiteboard
Physikalisches System
Wechselsprung
Multiplikation
Unternehmensmodell
Gleitendes Mittel
Gerade
Leistung <Physik>
Verschiebungsoperator
Softwaretest
Kreisfläche
Mathematisierung
Prozessautomation
Groupware
Physikalisches System
Quellcode
Frequenz
Integral
Software
Rechter Winkel
ATM
Mereologie
Basisvektor
Leistung <Physik>
Zentralisator
Impuls
Bit
Demo <Programm>
Umsetzung <Informatik>
Prozess <Physik>
Punkt
Mengentheoretische Topologie
Gruppenkeim
Iteration
Bridge <Kommunikationstechnik>
Kartesische Koordinaten
Gerichteter Graph
Formale Semantik
Rechenzentrum
Netzwerktopologie
Softwaretest
Datenmanagement
Web Services
Gruppentheorie
Prozess <Informatik>
Code
Computersicherheit
Skript <Programm>
Elektronischer Programmführer
Figurierte Zahl
Auswahlaxiom
Gerade
Prototyping
Metropolitan area network
Web Services
Softwaretest
Nichtlinearer Operator
Schlüsselverwaltung
Computersicherheit
Gebäude <Mathematik>
Prognostik
Programmierumgebung
Bitrate
Biprodukt
Frequenz
Teilmenge
Arithmetisches Mittel
Kollaboration <Informatik>
Dienst <Informatik>
Diskrete-Elemente-Methode
Menge
Rechter Winkel
Beweistheorie
Projektive Ebene
Ordnung <Mathematik>
Repository <Informatik>
Programmierumgebung
Schlüsselverwaltung
Prozessautomation
Beweistheorie
Geschwindigkeit
Subtraktion
Quader
Selbst organisierendes System
Mathematisierung
Ablöseblase
Automatische Handlungsplanung
Gebäude <Mathematik>
Unrundheit
Dienst <Informatik>
Whiteboard
Code
Physikalisches System
Virtuelle Maschine
Reelle Zahl
Mittelwert
Datennetz
Speicher <Informatik>
Soundverarbeitung
Videospiel
Graph
Betafunktion
Einfache Genauigkeit
Integral
Dreiecksfreier Graph
Ruhmasse
Visualisierung
Unternehmensarchitektur
Data Mining
Bit
Punkt
Kontrollstruktur
Natürliche Zahl
Mathematisierung
Versionsverwaltung
Automatische Handlungsplanung
Abgeschlossene Menge
Maßerweiterung
Code
Whiteboard
Physikalisches System
Open Source
Zufallszahlen
Lesezeichen <Internet>
Code
Punkt
Operations Research
Softwareentwickler
Maßerweiterung
Druckertreiber
Figurierte Zahl
Mathematisierung
Einfache Genauigkeit
Routing
Physikalisches System
Quellcode
Biprodukt
Datenfluss
Integral
Rechter Winkel
Gamecontroller
Wort <Informatik>
Unternehmensarchitektur
Umsetzung <Informatik>
Abstimmung <Frequenz>
Punkt
Weg <Topologie>
Versionsverwaltung
Unternehmensmodell
Gesetz <Physik>
Login
Monster-Gruppe
Metadaten
Einheit <Mathematik>
Datenmanagement
Computersicherheit
Skript <Programm>
Metropolitan area network
Nichtlinearer Operator
Computersicherheit
Gebäude <Mathematik>
Ausnahmebehandlung
Programmierumgebung
Quellcode
Biprodukt
Bitrate
Kontextbezogenes System
Kugelkappe
Teilmenge
Kollaboration <Informatik>
Menge
Wurzel <Mathematik>
Rechter Winkel
Beweistheorie
Server
Garbentheorie
Projektive Ebene
Versionsverwaltung
Programmierumgebung
Fehlermeldung
Proxy Server
Server
Wellenpaket
Selbst organisierendes System
Mathematisierung
Zahlenbereich
Gauß-Funktion
Netzadresse
Open Source
Systemprogrammierung
Multiplikation
Knotenmenge
Domain-Name
Datensatz
Datentyp
Monster-Gruppe
Softwareentwickler
Normalvektor
Digitales Zertifikat
Physikalisches System
Binder <Informatik>
Elektronische Publikation
Quick-Sort
Mereologie
Gamecontroller
Unternehmensarchitektur

Metadaten

Formale Metadaten

Titel Introducing Delivery in Enterprises - Lessons, Techniques, Tips and Tricks
Serientitel ChefConf 2016
Autor Oakley, Caedman
Lizenz CC-Namensnennung - 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 10.5446/34638
Herausgeber Confreaks, LLC
Erscheinungsjahr 2016
Sprache Englisch

Inhaltliche Metadaten

Fachgebiet Informatik
Abstract Cædman has been working closely with Chef to introduce continuous delivery into an enterprise ecosystem. The company is using Chef Delivery within an existing Chef infrastructure, and has many of the normal issues that people run into: workflow changes, high-security requirements, and resistance to change. This talk will cover the journey from a bare, mostly sort-of CI, shop running a single Jenkins master to a company that has services running on Chef Delivery, and being continuously delivered. We will cover "Barriers to Adoption", "Solving the Soft Problems", "Delivery Truck? Oh Fudge", "The Road to Deployment Is Paved with Good Intentions" and other topics. At the end of this talk, you will be able to navigate the minefield of moving an organization from a traditional set of models to a more complete continuous delivery model. not just from a theoretical standpoint, but by seeing examples of what went well and what went badly.

Ähnliche Filme

Loading...
Feedback