Bestand wählen
Merken

Orchestration of Services with Juju

Zitierlink des Filmsegments
Embed Code

Automatisierte Medienanalyse

Beta
Erkannte Entitäten
Sprachtranskript
let me introduce myself my name is Charles Butler I'm also at lazy power on IIsi Twitter all those other great places that you can find the i in the 1 true original is power and I'm going to talk today about so the key players we have in the boon to the ecosystem for orchestrating clouds we've got to hear your basic or to server we've also got snappy Cornell which was recently introduced for the Internet of Things and other of such as with to touch some great stuff I we got mass in the far side but were not a talk about those there were a focus and you to do is a at duties service orchestration framework but it's so much more than that because at its core really is an ecosystem where bind you you you're buying into a full ecosystem of tools and our community members documentation and so much more than that but before really dive into the different classes so that let's talk about what the motivation behind you was why it's important and why he was what what we got to this so in the beginning there was a single servers you would have somebody to working in the basement somewhere is the the stereotypes and you'd call them up whatever the e-mail server when done right mn e-mails down or no need servers on a scale so we start adding to these with clusters and this was still very much manual management and then there was the invention of the fertilization we discovered a wicked slice these machines get more use out of the servers by but throwing virtualization on top of a wicked man is these images and it was really interesting but the today we must not containers so when we start looking at the full stack and how crazy this's there's a lot of different moving parts to this and a gets increasingly complex to work and every different layer of that so let's throw on top of that the complexity of development you've got this guy who looks inwards that guy this guy's like nothing like either of those so we want a tool the work and everything yeah so once
again was original we want to effectively
deliver and maintain applications and by
doing that race but this down into 3 separate areas retired whether development the deployment and the outside of it so when start developing your infrastructure looking at infrastructures code you need a method that is consistent it's going to be that have high production parity you model year planets is close to what you running in production otherwise you basically testing something that doesn't necessarily represent what you will be working with an while and that introduces new variables that you have on the outside your control we also need these environments be shareable so if you're working on a component and it's got 3 services attached to it he bill share that with another developer and any bills sped up the exact same configuration and get moving with the quickly the we also need these to be readily available on the sincerity in the onboarding process can be very difficult it's often time-consuming and not very cost-effective so with all of these in mind when the the next stage of deployment so deployments 3 straightforward which was started in figure some services we also wanna put our applications there the water with the arts the we ask you to think about the service life cycle what is the servicer how are we gonna manage their service life cycle the how we can reconfigure services as we start introducing micro-services in the in the stack in the model and how we gonna work with the interactions between the services so do you has a goal
of encapsulating ops knowledge in any environment whether it's a public or private cloud the we also provide a dynamic infrastructure with interchangeable services so effectively turned your code infrastructures gotten interchangeable parts and turn it's like a factory line and the big when behind this is so that you can focus on being an expert in using your services and not knowing how to deploy it and how to run the service lifecycle management of the services that were not saying that you should not read the book and understand how to work with that service were say we were removed that 1st step we want this to be approachable in today's society with start-ups you're moving from idea to production is the editor is going to determine your chances to win if somebody else can do it faster than you they're going to get there 1st there and get the credit so but back and that's a 1
a citizenship card what I personal heroes genuine drives to come from he made a statement that resounded with me With such amazing applications he stated that nobody's playing Anjanette's or my sequel were playing legal legends and that's me just it like hit in the face of a ton of bricks because when you really stop and think about it when you keep focusing on the machines and thinking about the services underneath it you looking at 1 small part of that picture of you to take a step back and take a look at what your providing what is important you what you giving your customers what is pretty among what is keeping things moving you know what is what is keeping your community movie is project and that's
what binds to me so we all know the modern architectures service-centric because we started talking about micro-services containerization immutable infrastructure these are y and that and we have a model for that called charts but charms are more than just services charms actually encapsulate I and drink observe the Johnson encapsulate obs knowledge but they also dictate how it's going to work and interface with other services in a stack while some configuration management tool chains do work with the opposing services and they get into a little more petty things like service discovery and order discoveries we have defined a model the that makes this a declarative syntax and this dictates how your application will reconfigure given any
as specific configuration so not only can you configure service with like you know I know the 1 have 80 % my ran dedicated to my sequel we can also state that once we related to a slave that that slave is not to be re rapid app so
with a gap in big management is the top level which is often times a lot of people get high happening in LGG was not a configuration management tool we actually set a layer above and stack for doing orchestration
the reason I SS-CoSBI language-independent we want you to bring the skills that you have with all of your other tools and is rich ecosystem and continue to make a cost and you can use whatever tool using today even if that's bash the chef puppet interval salts that I'm sure that we got so many representing a tool that you know nobody else is really heard the paper I worked on themselves but there and on
top of that is also operational intelligence the reason being I sort of thought about this little sidetracked the relationships which is the line between MediaWiki in reading the is how word is how we're doing the orchestration this is how dictating things up in how the events system works in the environment so we can change up Maria with my sequel we want to how does that work the which are looking at here is a is a yellow file and this is metadata diagonal this is the 1 core Follett's required to make it due to chance everything else is optional as long as you have this you have made a chance it's like a doing things you know what would be the point here environment but at least it will understand how to work with it so it's got a name summary 1 tell us who the maintainer is the weakest things break the user can get back to intact maintainer gives a description to redisplay this in something natural a little later we target for organize categorization and then the big important part is right here research requires this is where were dictating to dude you was going to talk to our service these are the things that I understand I understand the the understand slave owners think cash and what's cool about that is that there's 2 components this not only give the relationship name you also have what's called interface interfaces where the magic happens this is where data is going to be exchanged across the wire so you can have any number of relationships that share but as long as they have their own unique interface that's we can start stacking different services together using the same relationship and getting that same data across wire which enables these interchangeable components provides is something that your charm going to expose to the environment so viewpoint a web service and it provides a website relationship that means that certain assigned over the private address and probably port 80 so that we can go to a load balancer so interfaces
are loosely typed contracts meaning that we don't superimpose any kind of a Nazi there is no standardization to it you are free this is gonna while West and you'd where you can get in there and really sort of private and start defining your applications and how they talk to 1 another so given the example on MediaWiki so I know how to connect databases that implement my sequel manner where I try to relate this money DVDs afterward right because money DVDs is not impolite my
sequel so best with the general we see that requires my sequel we don't have my sequel anywhere on Monday B so this relationship will not work is not exposed if you try to run the command line going a lot traded to the gooey it's complain that you actually won't even let you drive but so we know that this is in our now we're looking Marie DB where requires TV interface my sequel really be provides interface my sequel that's a match gossip can work so the touch a little bit more what I was saying
earlier about getting to market faster working with the application getting your ideas and production I'll write size of a gentle learn the rules of the game then have to play better than anyone else an apple at this point with a declared a model with got pre-cluster that because orchestration really is agility a logit interchanger components and test ideas and move between these ideas how many of you seem you deploy OpenStack anybody a the a couple Jews you is the underlying engine that empowers input to deployed OpenStack medicine out a pilot service were really mark will bring somebody on stage so here please because services you want click start but the OpenStack at the very complex deployment to doing it by hand it can take you a couple days to a couple weeks to understand the process of what services need to be configured how they fit together it really is an operating system for the web so looking at that if you're a start the point of the fact that I have all the services understand another together what information is sent over the wire how these applications should be configured you Mr. Figure 1 of those components this is basis I'm configuration management cited but then where we get to the newer services and will start changing things out it's amazing due enables that this is a big data stack this is working with Hortonworks to do and it's hooked into a logging system with all that stuff friend by elastic searching cabana these are all available and we let's call them executable white paper because were encapsulating the best of the community's knowledge distilled into these charges and we distribute them through a charm store and that undergoes peer-review it was beautiful about that is that it's bundle happiness bundles define your entire infrastructure in single file they can define components or they can define the entirety of the infrastructure so the development you know the services together employed shippers soccer stadium redeploy run tests against it is a great you to production same thing high production parity it happiness across the board
so in this model development not really work together developers get to the play applications
they have production parity and get the opportunity to define the core infrastructure that you developers work on they can also ordered anything that's coming in for security and they can also operate at scale and developers allowed the ship right into the infrastructure so often do the review so looking at this from a business standpoint developers get to work with only the concerns and ops get the ability to modify the entire bundle and they can move services between clouds quickly so if we look at this is decomposed In
a typical environment you might have your physical layer infrastructures service which is you just important to servers for OpenStack above that you can delegate concerns with the V ends in the containers were the desk at point and work with their so that the site infrastructure and this separation of concerns really helps it to move at the speed of thought because with this decomposed infrastructure you got your core service with your time and the developers now free to experiment and work in move amongst the different thing is in the hands so
you might be asking where you can use to you were in every public cloud that is listed here with that in the web services giant HP when is your who compute engine and digital ocean as a couple but I haven't mentioned because using the menu provider you can use any clever that's public or private you lose a little bit of a magic which is investment or if you're using it with Masterkey orchestrate with their metal so if you guessed service layer you can just let this together start applying today so let's talk about the 2 and we have to facilitate your creativity dude you do is arguably the best implying here because I like to call it manager mode the really it's introspection this is a you can visualize everything you're working with in your environment in 1 go each and every 1 of these charms represent a service that you have deployed and you can see all the relationships between the different services so if you were to take vacations and some we were to come in and say hey what's going on there is a red line here the between Mozambique's agent and dashing lizard there's an error it's going out the configuration and we wish to change that was back the results and you
quick-start which is you help you get up and moving quickly it's gonna bootstrap environment deploy the GG do it and expose it for you but this is the thing that gets me excited the duty charm store is the collection of all of the charmer recommended charms this is including our OpenStack charms obligated charms Apache the strands around again typically start by community members submitted the code review process and with an iterative process they have big after we've added tests we run testing deployment on every single cloud all these terms now under CI so this is certified to work in your structure and a cloud the you're getting the benefit of everybody that is use these charts and contributed back if already a label I would say it's a lot like a forge but this forges is open for everybody and every 1 of these charges license with OSI-approved license now some of them may have MIT and Apache license word you feel 3 that's up to the author of the component here that's the only requirement we have is that they have testaments and also provides so if you have any
questions but what I just ran through really quick like I'll put my slides up online and now you can find on Twitter Ali's power and this will be put those where you can join is an icy different a net and you read a reader of the mailing list now I made mention the last talk that I was going to run a quick
experiment using some sentiment analysis
so I invite you to tweets at the Poseidon 15 hashtag n let me be the 1st 1 is say that
this is not a toy race and analysis and we will look at how this is put together in a point would you do I was gonna try redeployed Islamic stage since I don't have SSH access that's not gonna work but what's going on is that this application is trolling Twitter and the reason we get a new tweet the comes in with the hashtag of falls in 15 it's going to split that up this can plot the common words end of the it's gonna run the remaining letters or other words through its analysis was at so if you're is something like bottom 15 this is awesome with the lazy part stage self blood who and at that point is the run through the same page that were also as the rank it's we give a score and you'll see the green the greengage go up and there will be some of the false-positive seriously and knowledge of the idea of life they don't use this is like a legitimate this is the science behind said analysis now this is like the most basic thing that I could put together before but the demo so looking at this only this up a little bit I'm not going to invite you to ask questions before we dive back into the gooey and take a look at the plant environment and what we can introspect the are there any questions to the the now that's really love it you ask that but that's part of the key to building a rich environment is defining proper relationships interfaces so as long as the same information is required would say that you're using large stash on 1 machine where you can also use hacker shippers lots of this slot shipping agent now as long as it's require the same information just an IP address and maybe a porpoise disseminated to at that point you're going to go if you have multiple instances you can define that as being a leader or if you run into into a situation that requires a quorum you can run elections on the charm coated terminal here so that way it sentences consistent information back over the wire there are patterns that we have in libraries were developing the help with this note some of it is a little little if the only honest it the but there is you the 1 what of the you and which website is in right square OK so you using to separate my sequel server is in the related to 1 of the 2 uh sequel service correct OK that points the increasing against 1 database that's the relationship with if you define a relationship between the database and secondary database depends on how the Chalmers when reactive the Chalmers odd-even configured and receives the relationship that a 2nd time it's been Sally drop it histogram Uranus grazing the database now what I would suggest in recommend is that you have to set that up as a re replica and still a single database cluster unless you have the 1st secondary now we've given you enough power to do what you need to do to be creative we thus Paris yourself in the foot so use it's like pseudo you know experiment with that understand what's going on work with the tool but do have power to create catastrophic events cell is moderation any other questions assume here that the certainly i where even the place service you have an opportunity now I would love to do this right now but they said in which utterances state so it's this whole in my ability to do I demos but you can pass a tactic to and you can define KBM colon with the machine ID or you can provision on manager doing that and will go into a container or B M on that machine so it's can assume bare-metal by people find CGG deploy WordPress is going to employ WordPress to bare-metal machine or a cloud via given the environment but if I add a tactic to say lxy calling for its Winnipeg machine for the Lexie container this interplay WordPress inside of that yeah that is correct the charm itself does encapsulating the knowledge of the application the environment it's running in is the concern of the person supplying environment so but I'm not considered there's not terms the dog is we have a few leverage Stocker in obviously Dr. charm is going to play doctor container it would offer host it's not going to attempt to do this out of thin just in stuff on the anno is gonna try to fetch Stocker stolid pull the image down and they're right the it and that is correct OK juju is an event-driven system so as you I should apply something to the you it's gonna get a couple different states the 1st status right that is called install at that point you would do all of your preconfiguration and stuff like if you had a pen installing get maybe you're doing a get based delivery platform and after that can run into a configuration change took now With these events happening in sequence they're going to cues to say that you relate something what's the point once it's reached its stock market's going to go into the relationship changed so it would be really nice if my laptop and decide it was really sleep so
so with that in mind let me actually pull a charm up the and what's actually dissect
1 that only in a spike in tight which pull actually pull the darker term I like this 1 because we got a lot of insulin here so I'm going to treat the hopes directory and what you see here all the different so excellent implemented for for the doctor service but the way that these are working sequences that here the installer which is Israel preinstalled and business the convict change Turk which if you expose configuration options for your job these run every time the user changes so say for example say dude said Dr. latest equals true which is going to install doctor from the upstream host right now stricken ecotourists states equipped that we the and you define these configuration options for your services 340 L format you gotta at a party of options and after that you start listing them out we support 3 present option types present these are kind of limiting were aware of this were actually working at war such as static lists so that you can kind of defined scope what users have the ability to choose but but the String boolean and numeric values of and working with these In the charm code the yeah we to see we actually open this to have step I can only see that too I
realize that I'm working on but I can see maybe now you can see better OK the this is actually Python wrapper and you'd so as I said you can it's language-independent you can bring whatever experience you have with you and we chose to do a a little bit of by Don shenanigans to make and will play really nice what you do In order work with that we've got this library culture Melva's and turnovers is going to bring in a few different niceties for people developing charts it's got automatically install interval for you it's also going to be a set of PPA and use the latest edition of we define some names that we are going to I expose so if you haven't name that is not in this it's not going to get run which is a nice way us to kind of what's coming and looking at this we define a playbook pass and then we also set the public's forward pass and so these are words and understand this is going to have the ability to runs you can implement your playbooks where you need to do on the host any sort all the ants wilderness will to you and you still get perform your orchestration for you so looking at this right here we've got a decorator for hook and celebrate charm this is where we actually run the magic of installing instable and it's got a she to run so not really exciting but this is the kind of glue they were using to make this chart compelling so now we're assuming playbooks we've adopted a pattern of trying to keep or concerns are playbooks nice encapsulated so we we run install or upgrade there's only a couple things we care about whatever this the state occurs when action we got a doctor group on the host because let's see doctor and the doctor iupac both require this and we also add the user who to the doctor group so if you're familiar with danceable this is very straightforward the the this is actually an older copy of this term I but also at this point we wanna understand what's going on in the environment we wanna find out of pocket I was installed we register those and we have those variables available to us further down the call stack now at this point these are all driven so you gotta think about how your writing your charms in the sequence of I have an event is coming and I had the i've defined a desired state change to my environment know what actions are it's or is it going to take to reach the desired point and I hope I have got the tests in here how do so we expose a really cool told toolchain to do this called and an angular is our testing framework so I'm switching feed on it just a little bit so bear with me I this is going unit test style syntax through a lot shorter and the set up is where we go through it we actually deployed units to our our system so so we have a little bit before about service back and what is important now where were testing infrastructure base changes we don't want necessarily test inside the charm because that's going to be a unit start tests for integration testing is 1 which of that the environment we declare a state change want to run an assertion that that state change has occurred so I we say really give it some time to wait because will run this on every different cloud provider at this point but as a split up will raise a an error message and then we execute on the testing the doctor was installed so for testing the darkest that we make sure that the file is there use urban doctor that's not a reason exception when I make sure that we can return info for the runny duck environment we also want to actually install an image and run inside of doctors were pulling the BusyBox image and at that point we run a couple different assertions making sure that we can actually run Hello world through that we get that upward back and then if not it just says that we can't with container but the test fails at that point the merge would be approved so 1 of the things that I alluded to the main article that is that were actually economical running all these charms in CI against all the major cloud providers to you get that for free by buying and or a binary system if you want your chance in the charm store they're going go on either CI were and take the legwork to making sure that those remaining high-quality experience for everybody who wants to use these it further questions just the that you universe yes picture of all the way to do this each with 1 of the the some all of the this is the configure IBM that I showed is actually expose on the host as a as an API driven command you got conveyed get which will then allow you to pull in the information that you said in the configuration I am also if you the whole warehouse something outside of your camping management call in there that would be the method that is defined in the configurable and polar back and or if it's information is review provided by relationship so the in of the of the view that the school this and that's correct that's crack so it crosses all those different barriers that we actually a few charms that implement both batch and Python just pure languages and we see this being intermixed and that's for a lot of people getting the story is a some of the uses this started working which you were only familiar with provisioning machine fashion we did 1 superimpose limitation and yet on Ruby or yet understand DSL we want people to get moving quickly with the system we want to get out of the way and this was 1 of the ways we're able to do that was by exposing this language independence and is giving the orchestration blue to kind benefit these components together and so on no questions just the the you about you could and I go a step further with this and state that In the Juju model you're buying into the GG state Server This is a client-server application we do have an agent get the stalls all written goats if its high polite way but if you're going to use something like chess server don't use due or moved using chefs in studio so you don't have to units that are going to do define state on a given machine and have them fighting over 1 another what a change occurs as I have seen this I actually play this game it's called a look at it 1st into the logs blow up an obvious it's really matter you so but but yeah that's that's just 1 caveat that you definitely made aware of that if you wanna buy into this were work with it it's perfectly fine to isolate the environments and have them work with 1 another but don't try to have 1 machine provision enable because you're on the problem of race conditions it the other questions who knows some analysis anything positive area 4
. 3 and a positive scale run on church you know the whole story under mean and couple of of on the amount of
you can but the question was a we have an OpenStack bundle you want to
know if he could move between the different kinds of OpenStack distributions like icehouse Havana et cetera you can the evolution of the OpenStack bundles is version so In order for you to go from say icehouse were from have and I cells it is you would you define the bundle revision that actually deployed have and now this is gonna be an evolutionary process with the bundles we provide because these really the best practice deployment of getting everything all at once so now you you feel free to take these and hacked them up and change them but sigh if I remember correctly actually
there is a core cool tool at due to chance
iconic you click on demo you're going to get a
disconnected gooey that you can interface with and stop piecing other ever component of your infrastructure so looking at a glance I do believe a glance exposes a distribution that last string so I'm going to double check that and all right it looks like I am incorrect in US but to what I can do because I'm almost positive you can do this I could feel I pull a blank I can get over OpenStack developers and make sure that that is the case that you can move between them it now that that is the power where or can't of minimizing the are common misconception we've actually got support landing thanks to cloud-based solutions always start deploying words for the point what is a mass Judy was not coming to orchestrating windows as well but we get a community member this working importing this the CentOS the only thing that's required to actually interface would you the core client is you have to have clout in and a pass user those are the 2 things that makes assumptions on the outside of that the charms now are going to be the impediment if you wanna take the existing charms and move it from a Debian-based Maupin based system I said most of those while I say merciless just make a broad statement of all of them are since right now the geared towards a orchestrating boon into but there's nothing stopping you from taking a chance an action it can work with European-based stress yeah the that is planned I I wanna say that is on this upcoming sprint but I I hesitate to give you that is already TA but it is defined on the road map now as far as wondering we're charms that expose that we wanna bring some more that logic back in the core and the weights integrated you but is it is kind of a bummer to say halves deployed 50 machines would you you would Osterman and you get an obvious and look at all these 50 machines that having a national health check it like a everything is called the readability checks everything's paying fine as but what's interesting is that the charms are going to encapsulate what is going to be the service for the health of your service so have a hope between around every so often that I like to check the state of the machine in the environment and run reports on that as well I'm sorry little that's hard for me say I think that the 1st step is just to get a reporting and after that if we everyone add more large to that such as auto-scaling but that certainly in the world possibility but it's not something that would necessary on the about today just the period of loss of all that so you have to use that right to OK of the in the the correct but that the change that you make your 1 export as a nail file because that's going to be that that the perfectly your deployment environment for every change you make whether that's configuration change for upgrading aversion and that's what you would be CS is the bundle of your environment now the gooey is very much a 1st class citizen because it's communicating directly with the State's server the state servers writing this or that commodity and longer DB is actually managing those and we've got us so my transactional state right now using the going we don't have a to where you can roll back but you can stage changes before you commit so secure transactions are coming I wanna say there's going to be the end of this so the cycle but as landed yet so nor details on that story the but there's only the 1 verse alive in your environment but as the bundle itself the iterates in the store as the services get upgraded those the options available to you in the gooey environment or on the command line you can and you upgrade charm it'll pulled on the latest version not like an overriding configuration URI specifies as an upgrade in place so say for example How would you would see thing the came out and say that you've deployed the Ubuntu Server chart so whatever the next 1 comes out and says hey this is real feed upgrade stages you upgrade would to and that was going around Paulie rapid updates as confederating Fourier and new questions check of the as far as always upgrades we version all the chance with tied to a series and we typically recommend people use LTS for the deployments so that we modeled after we're gonna precise a is the trustee series and I believe that it is the next LTS moving actually in Part for shops we find our you know I have an X that anybody asked questions as a really good 1 I know that you would have to deploy the trustee series of charm I don't believe you can upgrade so you to put a precise version of WordPress and then trustee comes out I believe the yet to deploy the trustee version of the WordPress drama to move to the next release they at that point data migration this independent you have the WordPress strong configured if you've got something is working around a fastener storing all the user dataset for their new after relate that to them charm at that point consolidate over you may have to find some of the username password stuff because that's generated based on the charm ID but as far as the start of that what you already deployed it without you date is there you might have to do a little bit of manual work on that such as the configuration of editing using passive the database but aside that should be fine the we story is something that's also on the roadmap for being implementation but uh we have a block storage broker charm that talks with OpenStack based persistent storage as well as the AWS EBS volumes that so that we don't have support so we're looking at that and core for the different cloud providers but not all them expose that presented a pointed allusion you don't have that option in the Options with an NFS server for mapping into some third-party service so little tricky there were looking at we've already started to address it with charms rustling in addressing it in court if the other questions the that can be used for this kind stage of of the and here you can find in a way of major right and you call this a little more about the I would say that it is fast taking your existing infrastructure and working in which you do that you would start transitive transposing that 1 service time I know that's not the greatest dancer were already looking at other ways to to transpose of that's actually the tire somatic configuration injury camp is how we can address this issue moving forward gonna across info with what people would like to do with that we wish for things such as using my consul to start probing increases Asia that you can store near the services and making aware so would you can they communicate with them and talk to those other services but as a stands that AGG was only aware of the services that you always Our through work with other services you can use things such as proxy terms like there is a need of USS 3 plug-in to given ephemeral storage and just since the credentials that you can use the back of a wire to your application and the consume those but outside of that you know there's not really a great story and and is what old the of the could you repeat that for me do I have to go all the of the only is of the order of the OK I we don't have scaling that's actually tied in with our Madison health reporting that we're talking about coming up this upcoming cycle of whatever that does land the idea behind it is you deploying orchestrator charm that can communicate with the environment and then based on the self checks it would then make the actions based on things you define now if you take a look at a Cloud Foundry chance we got orchestrated you exodus deploy 1 Charlie environment you pick the verse of CloudFoundry that you want to build your past for you and deploys it in your cloud providers at that point it's also you take a look at it and also handles with a reconciler to do any kind of scaling actions you might need to do on that service so that's the kind of intelligence overlooking the vacant but the question is does a blonde and Chalmers a belonging core so we're still having a conversation about that and I we have to do by you furious that to join the conversation the the the the the the right right the thing is that it is it's gonna varied depending on the person as well as the clouds so if you've got a hard limits the you so this thresholds for your scale like maybe your painters of the 5 seconds you don't want scale because at that point you know you've got you know maybe an extra 15 20 people and then also year flood comes in and then you get a hard scalable you you 30 second response time so then your automatically insert scaling right because now you've had the threshold so that would you define those and would you want the logical right in the moon see you only
taken my history the you
have you are the stability the cinnamon analysis stack we actually employed I also discovered that you can connect directly IP over this Wi-Fi which was interesting all all 1 faster so now what I wish well and then you like before it I took the stage I think that is choice info OK let me use clients and the a the hurricane on duty and work so as I P V 6 verses
ideally foreigners call nice call the OK so looking at this terminal the output of what I get back from duty status is that you return in machine readable in human-readable Gamal format but this particular service takes up but see 12 years but I think it's closer than 9 I necessary
Our this does a lot the greatest on this resolutions on a try to step through this as carefully like on but when it was that I had the sidebar there's a there's a whole slew keyboard shortcuts appealed shift in press the question mark
you can to the going to do interesting things such as renaming environment so I feel like you
need a you as you can rename this to add 0 you know
flies from rocks I will say that and then
only for this crew recession this year this is known as far as rocks a so that's gonna call for like demos and getting stuff way but this control on the side is actually a look at your different units so Kafka as a service that I've I've
got that what would put together was a land architecture there's no tackled Kafka Twitter and what it's doing is it's trolling the Twitter API pulling streets and it is the the Kafka happens what some a is what's responsible for splitting is up and pulling out the the normal words and then at which point ship that information of 2 Apache Storm words in a convertible spot job In a star and actress analysis so the Juju cited this as I can see that got 1 unit of copper running I don't need more than that it's got no configuration and sets several relations available it's running the duties on East Asian HDP zookeeper and it's also related the copper Twitter Kapitalismus alluring application so take a look at a storm workers the several here so this is looks slightly more compelling of screen we could see all the different machines we got making this up but they do we still retains the representation of just the service that we care about I click on this I can see all the information about about it I've got a public IP address nothing is currently exposed me that I can reach it over the web is only available over the private IP address so still behind the firewall as in a given a listing of the relations you see what else is important about this the configuration Snow configuration the so but we also have what's called a machine view which allows you to do direct placement such as you're asking before about how you deploy to a specific machine I haven't tactic to the Dewey also offers this using I using this pure here so for example if I click on unit 24 and an hour I need a call shenanigans in AWS I
us to change surge if you enable container control quixote a case where
a click in the machines and I condemn here at Unit
24 for which is why zookeeper unit on the click Add a container so as which type a when he's when he's Lexia Katie I go ahead and provision Alexi container and at which point it is staged scatter blue icon next to a tiny that have state something it is helpful matches the pockets as a you've only done something in staging you need Committee environment for x happens so if I
click on that go and say that we wanna add this here and now this topology has started adding and provision that Lexie container so back to what I was trying to earlier is something wasn't in Fig 1 iteration the doesn't have my Twitter API credentials but that here we are so the Sun Analysis Unit just exposes a port from 48 thousand this is another that sexy receiving the data being emitted by storm and if I were to change this from a thousand US 80 80 it would stop the service reconfigured undergo and break it is were most of in my talk so why not and to confirm and at which
point where we pull up the 7 analysis this I frame should fail
to start learning alright there we go it's like having a with the do rooms like I got the cyclic and page here notes
so it reconfigure itself it's not running on port 80 80 stories emitting messages know where I have just introduced a catastrophic problem with my environment so this is what I'm saying by we expose enough opportunities yourself in the foot but hopefully you've got something modeling such is not just whatever your system checks are so we get health checks built into due due
but moving this back to pour 8 thousand this should resolved a problem tha
can a couple seconds well I may have broken
the and I didn't the I was hoping right so if there's no further questions this this is really all that I've got I don't have the time to run through a full deployment will take another from 12 15 minutes just to get started so so are there any further questions regarding due due no OK my slides ago online as at power on twitter and I will be
running workshops at configuration in the continent symmetries of of the deuteron consuming
will talk about term but and thank you
thank you to the me the
Virtualisierung
Program Slicing
Kolmogorov-Komplexität
Dienst <Informatik>
Komplex <Algebra>
Framework <Informatik>
Virtuelle Maschine
Datenmanagement
Cluster <Rechnernetz>
Softwareentwickler
E-Mail
Bildgebendes Verfahren
Metropolitan area network
Leistung <Physik>
Streuungsdiagramm
Schnelltaste
Zentrische Streckung
Ruhmasse
Systemaufruf
Internet der Dinge
Fokalpunkt
Dienst <Informatik>
Twitter <Softwareplattform>
Mereologie
Server
Speicherabzug
Streuungsdiagramm
Streuungsdiagramm
Videospiel
Server
Softwareentwickler
Prozess <Physik>
Gemeinsamer Speicher
Wasserdampftafel
Virtuelle Maschine
Interaktives Fernsehen
Kartesische Koordinaten
Dienst <Informatik>
Biprodukt
Gruppenoperation
Variable
Dienst <Informatik>
Informationsmodellierung
Flächeninhalt
Gerade Zahl
Dreiecksfreier Graph
Gamecontroller
Gerade Zahl
Zusammenhängender Graph
Softwareentwickler
Figurierte Zahl
Datenmanagement
Kartesische Koordinaten
Fortsetzung <Mathematik>
Dienst <Informatik>
Code
Virtuelle Maschine
Datenmanagement
Gerade
Expertensystem
Streuungsdiagramm
Expertensystem
Nichtlinearer Operator
Befehl <Informatik>
Diskretes System
Spieltheorie
Gasströmung
Programmierumgebung
Biprodukt
Chipkarte
Texteditor
Dienst <Informatik>
Mereologie
Faktor <Algebra>
Projektive Ebene
Geheimnisprinzip
Programmierumgebung
Streuungsdiagramm
Streuungsdiagramm
App <Programm>
Dienst <Informatik>
Informationsmodellierung
Konfigurationsverwaltung
Verkettung <Informatik>
Fortsetzung <Mathematik>
Kartesische Koordinaten
Dienst <Informatik>
Ordnung <Mathematik>
Konfigurationsraum
Streuungsdiagramm
Konfigurationsverwaltung
Datenmanagement
Konfigurationsraum
Datenmanagement
Keller <Informatik>
Dienst <Informatik>
Übergang
Schnittstelle
Kategorizität
Punkt
Desintegration <Mathematik>
Adressraum
Zahlenbereich
Kartesische Koordinaten
Fortsetzung <Mathematik>
Dienst <Informatik>
Datenhaltung
Lastteilung
Deskriptive Statistik
Web Services
Zusammenhängender Graph
Gerade
Implementierung
Schnittstelle
Streuungsdiagramm
Datenhaltung
Physikalisches System
Elektronische Publikation
Ereignishorizont
Quick-Sort
Design by Contract
Softwarewartung
Dienst <Informatik>
Diskrete-Elemente-Methode
Mereologie
Wort <Informatik>
Speicherabzug
Diagonale <Geometrie>
Programmierumgebung
Standardabweichung
Maschinenschreiben
Bit
Punkt
Prozess <Physik>
Schreiben <Datenverarbeitung>
Fortsetzung <Mathematik>
Kartesische Koordinaten
Dienst <Informatik>
Whiteboard
Informationsmodellierung
Benutzerbeteiligung
Spieltheorie
Netzbetriebssystem
Zusammenhängender Graph
Softwareentwickler
Speicher <Informatik>
Figurierte Zahl
Ganze Funktion
Schnittstelle
Streuungsdiagramm
Softwaretest
Matching <Graphentheorie>
Spieltheorie
Einfache Genauigkeit
Schlussregel
Physikalisches System
Ein-Ausgabe
Biprodukt
Elektronische Publikation
Konfigurationsverwaltung
Dienst <Informatik>
Gerade Zahl
Basisvektor
Information
Faserbündel
Streuungsdiagramm
Nichtlinearer Operator
Zentrische Streckung
Softwareentwickler
Computersicherheit
Kartesische Koordinaten
Dienst <Informatik>
Biprodukt
Informationsmodellierung
Dienst <Informatik>
Gerade Zahl
Speicherabzug
Speicherabzug
Softwareentwickler
Faserbündel
Streuungsdiagramm
Streuungsdiagramm
Resultante
Trennungsaxiom
ATM
Web Site
Bit
Punkt
Dienst <Informatik>
Service provider
Service provider
Dienst <Informatik>
Web Services
Datenmanagement
Digitalisierer
Server
Speicherabzug
Softwareentwickler
Konfigurationsraum
Programmierumgebung
Gerade
Streuungsdiagramm
Meta-Tag
Fehlermeldung
Streuungsdiagramm
Softwaretest
Autorisierung
Gewicht <Mathematik>
Prozess <Physik>
Bootstrap-Aggregation
Mailing-Liste
Dienst <Informatik>
Term
Code
Rechenschieber
Twitter <Softwareplattform>
Wort <Informatik>
Zusammenhängender Graph
Speicher <Informatik>
Datenstruktur
E-Mail
Programmierumgebung
Streuungsdiagramm
Leistung <Physik>
Bit
Demo <Programm>
Punkt
Kartesische Koordinaten
Fortsetzung <Mathematik>
Homepage
Datenmanagement
Mustersprache
Radikal <Mathematik>
Hacker
Datenhaltung
Katastrophentheorie
Ereignishorizont
Dienst <Informatik>
Histogramm
Twitter <Softwareplattform>
Rechter Winkel
Server
Information
Programmierumgebung
Schlüsselverwaltung
Aggregatzustand
Instantiierung
Sichtbarkeitsverfahren
Web Site
Subtraktion
Folge <Mathematik>
Mathematisierung
Zellularer Automat
Dienst <Informatik>
Term
Systemplattform
Netzadresse
Virtuelle Maschine
Modul <Datentyp>
Webforum
Rangstatistik
Programmbibliothek
Konfigurationsraum
Hilfesystem
Bildgebendes Verfahren
Widerspruchsfreiheit
Leistung <Physik>
Analysis
Streuungsdiagramm
Videospiel
Finite-Elemente-Methode
Einfache Genauigkeit
Physikalisches System
Quadratzahl
Mereologie
Wort <Informatik>
Streuungsdiagramm
Streuungsdiagramm
Folge <Mathematik>
Mathematisierung
Zahlenbereich
Mailing-Liste
Dienst <Informatik>
Kombinatorische Gruppentheorie
Term
Code
Konfiguration <Informatik>
Hydrostatik
Dienst <Informatik>
Prozess <Informatik>
Datentyp
Dateiformat
Installation <Informatik>
Boolesche Algebra
Konfigurationsraum
Verzeichnisdienst
Aggregatzustand
Zeichenkette
Bit
Punkt
Komponententest
Formale Sprache
Gruppenkeim
Kartesische Koordinaten
Dokumentenverwaltungssystem
Login
Datenmanagement
Einheit <Mathematik>
Hook <Programmierung>
Mustersprache
Phasenumwandlung
Feuchteleitung
Softwaretest
Zentrische Streckung
Sichtenkonzept
Systemaufruf
Ausnahmebehandlung
Ereignishorizont
Arithmetisches Mittel
Dienst <Informatik>
Computerschach
Menge
Rechter Winkel
Konditionszahl
Heegaard-Zerlegung
Server
Information
Ordnung <Mathematik>
Programmierumgebung
Message-Passing
Fehlermeldung
Aggregatzustand
Folge <Mathematik>
Subtraktion
Ortsoperator
Gruppenoperation
Mathematisierung
Dienst <Informatik>
Term
Framework <Informatik>
Virtuelle Maschine
Variable
Informationsmodellierung
Spieltheorie
Wrapper <Programmierung>
Programmbibliothek
Inverser Limes
Zusammenhängender Graph
Speicher <Informatik>
Grundraum
Konfigurationsraum
Bildgebendes Verfahren
Analysis
Streuungsdiagramm
Stochastische Abhängigkeit
Physikalisches System
Elektronische Publikation
Cloud Computing
Quick-Sort
Integral
Flächeninhalt
Stapelverarbeitung
Streuungsdiagramm
Distributionstheorie
Subtraktion
Prozess <Physik>
Evolute
Versionsverwaltung
Zellularer Automat
Dienst <Informatik>
Dokumentenverwaltungssystem
Ordnung <Mathematik>
Faserbündel
Distributionstheorie
Bit
Demo <Programm>
Einfügungsdämpfung
Umsetzung <Informatik>
Punkt
Versionsverwaltung
Iteration
Kartesische Koordinaten
Dokumentenverwaltungssystem
Benutzeroberfläche
Softwarekonfigurationsverwaltung
Spezialrechner
Client
Konfigurationsdatenbank
Code
Bildschirmfenster
Addition
Serviceorientierte Architektur
DoS-Attacke
Zentrische Streckung
Befehl <Informatik>
Schwellwertverfahren
Transitionssystem
Datenhaltung
Machsches Prinzip
Reihe
Ruhmasse
Boolesche Algebra
p-Block
Frequenz
Mögliche-Welten-Semantik
Konfiguration <Informatik>
Software
Transaktionsverwaltung
Dienst <Informatik>
Lemma <Logik>
COM
Rechter Winkel
Ein-Ausgabe
Server
Information
Ordnung <Mathematik>
Normalspannung
Programmierumgebung
Faserbündel
Message-Passing
Zeichenkette
Aggregatzustand
Proxy Server
Server
Subtraktion
Gewicht <Mathematik>
Wort <Informatik>
Mathematisierung
Klasse <Mathematik>
Gruppenoperation
Implementierung
Dienst <Informatik>
Mathematische Logik
Term
Open Source
Quellcode
Virtuelle Maschine
Reelle Zahl
Migration <Informatik>
Inverser Limes
Passwort
Zusammenhängender Graph
Response-Zeit
Spezifisches Volumen
Speicher <Informatik>
Softwareentwickler
Konfigurationsraum
Disjunktive Normalform
Demo <Programm>
Gammafunktion
Leistung <Physik>
Streuungsdiagramm
Default
Konfigurationsraum
Zwei
Physikalisches System
Elektronische Publikation
Cloud Computing
Zeichenkette
Mapping <Computergraphik>
Uniforme Struktur
Mereologie
Dreiecksfreier Graph
Speicherabzug
Wort <Informatik>
Verkehrsinformation
Streuungsdiagramm
Streuungsdiagramm
Server
Client
Stabilitätstheorie <Logik>
Konfiguration <Informatik>
Code
Addition
Dienst <Informatik>
Information
Auswahlaxiom
Analysis
Streuungsdiagramm
Schnelltaste
Systemaufruf
Feldgleichung
Dienst <Informatik>
Quantisierung <Physik>
Virtuelle Maschine
Dienst <Informatik>
Radikal <Mathematik>
Dateiformat
Funktion <Mathematik>
Bildauflösung
Verschiebungsoperator
Demo <Programm>
Umsetzung <Informatik>
Punkt
Selbstrepräsentation
Gebäude <Mathematik>
Kartesische Koordinaten
Feldgleichung
Dienst <Informatik>
Information
Dokumentenverwaltungssystem
Netzadresse
Virtuelle Maschine
Benutzerbeteiligung
Einheit <Mathematik>
Prozess <Informatik>
Mehragentensystem
Endogene Variable
Konfigurationsraum
Touchscreen
Analysis
Streuungsdiagramm
Sichtenkonzept
Verhandlungs-Informationssystem
Relativitätstheorie
Systemaufruf
Mailing-Liste
Dienst <Informatik>
Twitter <Softwareplattform>
Heegaard-Zerlegung
Demoszene <Programmierung>
Gamecontroller
Ablöseblase
Wort <Informatik>
Binäre Relation
Computerarchitektur
Information
Programmierumgebung
Streuungsdiagramm
Punkt
Streuung
Iteration
Dienst <Informatik>
Bildschirmsymbol
Netzwerktopologie
Virtuelle Maschine
Dienst <Informatik>
Einheit <Mathematik>
Twitter <Softwareplattform>
Datentyp
Gamecontroller
Programmierumgebung
Aggregatzustand
Streuungsdiagramm
Tabelle <Informatik>
Punkt
Feldgleichung
Physikalisches System
Dienst <Informatik>
Dokumentenverwaltungssystem
Homepage
Informationsmodellierung
Magnettrommelspeicher
Programmierumgebung
Message-Passing
Analysis
Tabelle <Informatik>
Streuungsdiagramm
Rechenschieber
Lesezeichen <Internet>
Twitter <Softwareplattform>
Symmetrie
Zwei
Gateway
Dienst <Informatik>
Mini-Disc
Konfigurationsraum
Leistung <Physik>
Zeichenkette
Streuungsdiagramm
Server
Code
Dienst <Informatik>

Metadaten

Formale Metadaten

Titel Orchestration of Services with Juju
Untertitel How to orchestrate any machine or cloud with Juju
Alternativer Titel Configuration Management - Juju Orchestration
Serientitel FOSDEM 2015
Autor Ceppi, Marco
Lizenz CC-Namensnennung 2.0 Belgien:
Sie dürfen das Werk bzw. den Inhalt zu jedem legalen 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.
DOI 10.5446/34488
Herausgeber FOSDEM VZW
Erscheinungsjahr 2016
Sprache Englisch
Produktionsjahr 2015

Inhaltliche Metadaten

Fachgebiet Informatik

Ähnliche Filme

Loading...
Feedback