Paving the transition - Herding unicorns VMs on Kubernetes

Video thumbnail (Frame 0) Video thumbnail (Frame 987) Video thumbnail (Frame 2211) Video thumbnail (Frame 3113) Video thumbnail (Frame 9698) Video thumbnail (Frame 12468) Video thumbnail (Frame 15319) Video thumbnail (Frame 16328) Video thumbnail (Frame 17722) Video thumbnail (Frame 19492) Video thumbnail (Frame 21036) Video thumbnail (Frame 25234) Video thumbnail (Frame 26700) Video thumbnail (Frame 31995) Video thumbnail (Frame 33324) Video thumbnail (Frame 34704) Video thumbnail (Frame 43212) Video thumbnail (Frame 44246) Video thumbnail (Frame 46853) Video thumbnail (Frame 48523) Video thumbnail (Frame 51077) Video thumbnail (Frame 52417) Video thumbnail (Frame 54746) Video thumbnail (Frame 55735) Video thumbnail (Frame 58651) Video thumbnail (Frame 68963)
Video in TIB AV-Portal: Paving the transition - Herding unicorns VMs on Kubernetes

Formal Metadata

Paving the transition - Herding unicorns VMs on Kubernetes
Title of Series
CC Attribution 4.0 International:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Release Date

Content Metadata

Subject Area
After this session you should be able to have an idea of how VMs can be run on a Kubernetes cluster.
Keywords Persistent Migration

Related Material

The following resource is accompanying material for the video
Video is cited by the following resource
Theory of relativity Computer animation Internet forum Meeting/Interview
Human migration Trail Mathematics Focus (optics) Meeting/Interview Personal digital assistant Data center Virtual machine Point cloud Database Instance (computer science) Whiteboard
Point (geometry) Arithmetic mean Perfect group Computer animation Content (media) Virtual machine Video game Cuboid Lattice (order)
Classical physics Point (geometry) Multiplication sign Range (statistics) Port scanner Food energy Computer programming Medical imaging Term (mathematics) Different (Kate Ryan album) Software testing Damping Data structure Social class Task (computing) Authentication Electronic mailing list Content (media) Plastikkarte Database Cartesian coordinate system Substitute good Data management Arithmetic mean Word Computer animation Personal digital assistant Network topology Universe (mathematics) Video game Cycle (graph theory) Video game console Remote procedure call Quicksort
Constraint (mathematics) Touchscreen View (database) System administrator Software developer Multiplication sign Content (media) Virtual machine Virtualization Cartesian coordinate system Expected value Word Mathematics Process (computing) Computer animation Personal digital assistant Electronic visual display Physical system
Group action Direction (geometry) Software developer Content (media) Replication (computing) Cartesian coordinate system Product (business) Subset Word Computer animation Different (Kate Ryan album) Damping Endliche Modelltheorie Abstraction Tuple
Point (geometry) Group action Presentation of a group Service (economics) Direction (geometry) Data storage device Virtual machine Cartesian coordinate system Measurement Data management Computer animation Software Software testing Form (programming)
Greatest element Statistics Service (economics) Overhead (computing) Multiplication sign Surface Data storage device Virtual machine Set (mathematics) Instance (computer science) Mereology Theory Product (business) Human migration Data management Goodness of fit Computer animation Software Personal digital assistant Single-precision floating-point format Cuboid Series (mathematics)
Building Connectivity (graph theory) Multiplication sign View (database) Virtual machine Set (mathematics) Automatic differentiation Prime ideal Cuboid Damping Software testing Information security Computer architecture Area Data storage device Physicalism Planning Virtualization Storage area network Cartesian coordinate system Human migration Data management Kernel (computing) Computer animation Software Personal digital assistant Data center Statement (computer science) Game theory
Focus (optics) Projective plane Virtual machine Similarity (geometry) Bit Number Data management Voting Computer animation Different (Kate Ryan album) Natural number Energy level Software framework
Statistical hypothesis testing Multiplication sign Plotter Virtual machine Device driver Branch (computer science) Mereology Power (physics) Number Goodness of fit Mathematics Latent heat Peripheral Different (Kate Ryan album) Computer hardware Operating system Electronic visual display Stability theory Computer architecture Multiplication Interface (computing) Projective plane Content (media) Physicalism Volume (thermodynamics) Category of being Process (computing) Kernel (computing) Computer animation Software Vertex (graph theory)
Area Computer animation Different (Kate Ryan album) Connectivity (graph theory) Computer hardware Virtualization
Slide rule Type theory Loop (music) Computer animation Multiplication sign Virtual machine Form (programming)
Point (geometry) Implementation Group action Functional (mathematics) Context awareness State of matter Multiplication sign Plotter Execution unit Translation (relic) Mereology Field (computer science) Number Product (business) Impulse response Revision control Inference Goodness of fit Causality Different (Kate Ryan album) Operator (mathematics) Core dump Traffic reporting Computing platform Physical system Adventure game Constraint (mathematics) Namespace Fitness function Virtualization Price index Incidence algebra Limit (category theory) Cartesian coordinate system Connected space Human migration Data mining Arithmetic mean Process (computing) Computer animation Software Logic Personal digital assistant Right angle Game theory
Computer animation Demo (music) Cube Execution unit Instance (computer science) Spectrum (functional analysis)
Slide rule Game controller Demo (music) Multiplication sign Connectivity (graph theory) Device driver Function (mathematics) Connected space Medical imaging Voting Computer animation Software Different (Kate Ryan album) Personal digital assistant Internetworking Operator (mathematics) Phase transition Cloning Pattern language
Game controller Computer animation Logic Connectivity (graph theory) Mereology Proxy server
Human migration Type theory Word Group action Computer animation Personal digital assistant Cube Plotter Percolation theory Video game System call Arithmetic progression
Slide rule Pixel Multiplication sign Plotter Sheaf (mathematics) Virtual machine Regular graph Mereology Semantics (computer science) Product (business) Robotics Atomic number File system Constraint (mathematics) Demo (music) Block (periodic table) Data storage device Volume (thermodynamics) Arithmetic mean Computer animation Software Personal digital assistant Mixed reality output Video game Asynchronous Transfer Mode
Computer animation Demo (music) Cube Software testing
Classical physics Slide rule Interface (computing) Plotter Multiplication sign Virtualization Disk read-and-write head Cartesian coordinate system Connected space Data mining Computer animation Software Natural number Core dump Window
Group action Run time (program lifecycle phase) State of matter Multiplication sign Decision theory Virtual machine Mereology Graph coloring Rule of inference Usability Twitter Frequency Medical imaging Meeting/Interview Different (Kate Ryan album) Term (mathematics) Single-precision floating-point format Operator (mathematics) Core dump Energy level Diagram Software testing Information security Computer architecture Form (programming) Regulator gene File format Namespace Forcing (mathematics) Content (media) Lattice (order) Line (geometry) Cartesian coordinate system Radical (chemistry) Curvature Word Kernel (computing) Vector space Personal digital assistant Object (grammar) Arithmetic progression Window Active contour model Spacetime
Computer animation
yeah so the lesser see all of you know so dominated this forum watch and I'm going to speak about Cuba spot running but you machines on but in this
and I want to know over which is socialization solution for a couple of years and then we went over to work on Cuba and what people from all over the and start working on the working at red Linux company in the stuff so if we start there we can say that relations only presented pulp the
focus of this talk is obviously board was in the light of of this track to say no persistent migration so hard we couldn't get from today to to something in the future and but salacious only present today that means it's everywhere going um so getting cold I think this she had moved and so as they just on present day that means you have a change everywhere gotten classically in the data center with your own little machines to run the Oracle database we got in the cloud to GC I'm alone in Asia where you can create gazillions of instances with this little finger
but in this case it's really about about you or about us but we get our which machines over to the to the present day and it's not going today we just don't have the machines anymore but we also have containers they came up in the recent
years and I guess that you all know containers so you has used content for our life and who of you has used which machines before all that's perfect and so on container city as well the question is I if we look at the
than and we look at them and we see them and
we see all the heights and we see how much a stock of containers the so frantic tools with itself and the names and they look at the meeting smell a little bit like like which machines and so on but just better because it's new it's hyped can do everything you won't have any box mean and and then you to get to the point where we ask yourself so How can appear replaced by winds
was containers there are several reasons for that I
mean efficiency is 1 obvious point you can say that more administered on a host and gives to their quicker to launch and eventually easier to manage because they are text-driven have proper fire you can write their whatever you want to just rebuild the image if you want a new version so that be handed in Vienna Sweeney bigger provisioning tools we use and will sold the plates and OpenStack so continous are and we need to grant that Our really easy to handle so and so at some point you must all who has asked himself if you can replace you give content the no why did you ask yourself why you would replace them and so on I want to sort of structure and the it that you we're starting from scratch like a new car you don't want to take the old of you think and yeah so I think we're good on time so we can make it in this course so the thing is that we speak about replacement that we need to ask ourselves to there being a substitute teacher so too can contain actually replace the ends and term as set you look at the they look the same you can do some list of all the management tools the morals of trees some of of communities and the stocks for them and to I read 1 of these new tools from of what was called a like also something like that so there's a range of tools which which help managing but also brought the whole life cycle so there's really a lot of stuff from complaints and but then if you look in the detail what energy is used what impact does it have and for your application i mean in the end you don't run the ends for final appears before we run the because we want to deliver an application users to run at the end with the database to given that the is each or we run the and to give user and the desktop on demand so there's a reason why we do it from so what I'm why I'm saying that is because learning it where random them and we need to see they yeah if we can replace so we need to take a look at the technology and the to see here the containers can give us the same as the others so for example we speak about this of which university for example atom providing the users on demand like for other referring words or class of programming and task of so they might use of more test remote desktop solution that's just not available to so they can do obviously a serial console so but it's things you need to be aware of or if you got in at the hospital they use remote desktop solutions to prove their specific application for an scans or so remotely sometimes and use smart for authentication so in these classical legislation world you got features to do but passed through the smart card from you in terms of 2 so that it's not there yet in the world but on the other hand containers provide different features containers are easy to scale so if we look at communities the possibility to to be 1 container and scaling up to hundreds of what is not so easy with with the uhms on the other side so they are technical differences even if they look and feel the same and if they're really fit your use case then that really depends on the application of your support for the users so if you can replace that the
container really depends on the use OK
so in case you can't replace the a container with you then you will be with her containing all good so you might be job application you couldn't then that's really simple you
have to do anything well years later your you know that you have to do some stuff to really make it efficient and configure application to contain such the them but I want to focus more on the case we can say no I cannot not be replaced by the end of the because of the constraints I mean today virtualization would therefore out of 20 years or so different stages and the applications we have people in the room that system so you have the monolith this which are optimized to run the and they make assumptions about the operating system about all employ stuff so everything we as admins or with developers didn't last years was was built on these assumptions we have virtual machines like a lot of tools that help you why wise it's so there is the expectation that you have a screen displays that Y and you don't have you don't need that the requirement the content would just as well be considered wise words well up speaking what really we applications and surely that the major application changes due to the to the container technology but it takes time to change the if we think about complex applications on they business processes we need to change and the need to stop them and so you might say I could replace my view but I'll them together but I want want to do that and in the future when my application is ready for and good so now got
if you want to start now or if you want to start a future at least we have because we say we see transitions are see we are in the world today and we want to get started with and this introduction we I mean seriously we want get started the the questions to accept scratch with shiny new things and everything works but I think if you look at the contents themselves it up makes sense I mean you want to build up your containable structure of abstraction look at it but if you look at the application are using and the Kurdish into production OK so would you say the application you provide your customers of users would stop that application from scratch so you would be writing to fit into the country in the world yeah that's quite a lot of work so I think the infrastructure will be started from scratch tuples but there but the application itself that is not something you want to rewrite so we need to see how can we think we do we provide a replication customers because we make money and so and then we don't want to be but just for the reason we can infer new infrastructure so in reality we need to find a way how we can gracefully for all we we argue that this effect and we need to provide continued to provide all our product customers users and the question is how can we still do that how can we keep providing that application users even if we move or infrastructure for example in the direction of the status because of the benefits and they support different development models and H I was the word in that world and their faults and all this only words that were so it's not it's not that we want to write our nation from scratch but we rather want to see that we support and
evolution so that we take occupation and gracefully would into the direction to be able to run a business so we want to hold so how can we do that and that is that is slowly point where I want to get out of this presentation is how can we support the transition from the testing dataset novel today with the classic application the legacy you can call a legacy applications with today and into that world we have decomposed applications micro-services running on service measures and you can find another form more password you so the classical stack
very simplistic is that you have your infrastructure at the Beverly infrastructures storage and sound or not as dedicated you've got network switches and you got obviously you've got the physical hosts and they are aggregated into some kind of management things that virtual machine management applications the anywhere over OpenStack that many others just a few of them and then on top of them you've got the approach machines running which you need to keep running and continued to highlight that because we I mean I take and we live in reality that we need to see how we can really support our customers there's really was driving us all that we get from to take into the future we need to give them enablement and what support that so that's where we are today and then we say OK let's start from
scratch and bring up or a new management they which is optimized for contain sold we start again the that again we need service again and then we'll have a management they dedicated to continents and you see that on the right side of the Blue Books found but that means the twice infrastructure depending on what set up is the need to date the made storage device to this continent use case and you need to dedicate the network at all that whatever to that continues fixed and so you I mean if you would to start if you would you see something if you want get a feeling for it that's totally fine we speak about really the series migration after that you see on this work for me and somehow when the some time to migrate west of but no books and then in that case you probably don't want to make to make to production greater the infrastructure parts of the company actually before we you actually wanted making a single you wanted back up single us and you want to make sure that the single instance of switches along that don't have to so of double everything actually there's another set of what could do
you could say I take my existing the 2 machine set up and so but all of that I built my container management things so everything above the orange boxes that lots but the problem here is so that you still have to infrastructure layers because it got the real 1 at the bottom and the middle 1 we need what virtual storage so you need to have a VM which is serving storage and networking is not so problematic but they still need to manage it in the virtue machines so you still have an overhead because he's got 2 different layers we need quite the structure of for the changing of things so it's not I think it's good for the case and people to the theory of who did that we use butcher machines and they will contain as I read that's good for the statistics so I guess that's very convenient and I agree but again in practice it's that is to make it into to the surface so and that's what we get to Cuba
so Cuba tries to exactly going to that area so Cuba provides the
ability to run contains much machines on the same infrastructure so you don't have to provide 2 different 2 distinct set of all of all of of components for storage network and just physical also all the stuff in eventually due to data centers or 2 rooms physically depends really what we want to provide the customers so how does it look simple so here it allows you to
use the same management play that use the same physical storage in work Virtual storage network and other components you have in your in your new the and the management plane in this case is this communities but why is it that or why is good in our opinions because it allows you to me yeah the launch boxes so large because we have the assumption that the data of many butcher machines and start playing with contains but over to Prime you want to move stuff over where it makes sense so we set some applications be faster than others depending on the requirements some might not be able to to be mean move to every will think about security aspects somebody in the room is working security might know that unique against real complete a was isolated I want to go to the to put them in the details but containers these yeah they they effectively shared can go as we all know and the yams provide the ability to do not share the view on kernel podium which is good if you want to do some testing the kernel all for all the distance so that if you want different architectures when you on a single host and you still want to have the ads cost-cutting this is not a it's not a component which can give you the so yeah so you can slowly move in this like you've moved more war stuff from the butcher machines but the on the inside and the good thing is distortion network is not twice during the picture In reality that we need note statements to keep your side of the yeah and against some of them don't much different from a set you might want to keep some games where does make sense to move to come to this whenever you got questions drop me I think we've got time there regards but not that we looked at these Overview why will want to have Cuba because we look at it from the religious standpoint you cannot build everything from scratch again because it's all money it's all timing we can best so we wanted to see how we can provide a migration of 40 application so that that's a good living
and um that close look at about how this looks so this was pretty high level so far and I want to get a a little bit more in detail
now so common it is is the management they would build on on which is classically the the 10 framework to run containers on a common cluster on the number of votes but it doesn't provide the ability to find the changing there are different projects with Jupiter documents the very best you by far but we've got other projects like led and venture in nature something similar so they also run a virtual machines and and communities but with a different focus would all get to that in a minute so in very simple pictures if you now at Cuba to the picture then you're able to run the changing found all right so that actually is like this should
and so why stupid different than for example vertical rancher all who has the character of VerbNet rancher OK containers branches that are really interesting to see I thing renderers the project which is the closest to to Cuba by the way in and difference director and because of things content here then is no not not different stranger but in general or take is that we want to express the ends as they are so if you look at the ends and the containers or if you look into 1 of the communities introductions than if you look at the difference between the ends and plot you see that they have different properties and we spoke about that you can get span of architectures what what don't have the simplicity dictated by the architecture go below the answer can have some multiple place in parts don't have the graphical display at all is not something what what's the multiple networks Physical different devices you want to test device drivers to devise part through this is not something can simulate the pot you can for sure network interface to to plot but yeah it's not what do I it's a network device but it's not me but it's not really driven by a real driver it's not you know if you have that's power into those details which are difficult to explain them so don't really have device that for example if we look at the cages and the kernel in not physical devices their concepts inside the kernel to to connect and contains 2 to the hosts namespace them 1 thing that because we think that there are enough differences between ports and virtual machines that BMC their own definition so if no mistaken please correct me has been sometimes at least over that is 1 of the producers which is running in the nineties but the implicitly derived medium from the participation so therefore it felt that they have some work around the user annotations a number to specify the number of the but if you want to reduce the specific this all they use the volumes of what's to to attach this what setting displays and use the device that that's not possible because the derived from the that and the containers is doing the same so the polyp and derive their young from the from the pot but a set the parts that doesn't provide enough details and to to to to define a medium you can if we look at the operating systems from Microsoft Windows and then it's a fact that we need to have a stable hardware API so you devices should not change otherwise you are asked to register or re-authenticate on them to know the name of the process but we need to call Microsoft to get your all to to activate when again if the half devices change so we would really like to keep the yen API or ATI stable over time with venture over that of yeah even branches of the venture good that and click portraits of possible if you do operating humor changes below the API will change you can that happens in the activated course so we said we want to
expose the 80 so the virtual hardware completely as an entity communities and that's that's important that's the difference between Cuba and other solutions and so if you look at the the i in Cuba at the it then you will see and can you do that that you can
express from this really the immediate and that is necessary to keep that you may be unstable of who's familiar with liver eventually in so the British that such a component which has been around virtualization area of the most of the time and on we use it internally and that is why the in the
definition of the loop similar to the dominant of the liver and you also see that that actually the 1st bullet and so on that slide
is of 2nd bullet that we have a new resource type of instead that is what just saw so we've a specific dedicated implies we can specify many people so that will take some time to cover all the details because the problems and the details always think about the forms of a very good example and the ends of tuned for performance and then you get your money things like more than we need to make sure that so virtual processes tied by physical processes and if you do the price of through that divides us really aligned with the device node of the virtual CPU and that the stuff it is doesn't care about you they're working on it but we need the
tight control over because we speak what's the recipients we really want to provide production-ready Yemen's which you can use and support over over time and so what we can do is we can use this extensive here maybe I I just stupid but just show to to obtain use of the 2 so physical indicators the right so as to what the API the and 1 thing is that we want now go to the 1st bullet is that we provide everything imports and why do we do that
and if you look at adventure containers they all 3 of them currently you can run it on any communities clusters because it makes assumptions about the posts below so you need to install invertebrate case guessing that humans around it assumes that the relevant from what you've gone wrong so that stuff which if we look at communities of all clusters usually don't care about what operating systems below communities because you can run communities on in limits the number of dead in anoxic Fedora so the core was whatever and you usually don't care what is below you I want I have my container want to run it on the on the cluster so the new platforms a clustering of the host anymore which is good but In this existing solutions what we identified as problems need to rely on the underlying cause and that limits we can run the solution so we try to provide all the dependencies impulse so that can employed Cuba like any application communities you don't provide anything else actually we we try to be that good that we can act as a community that on so that the namespace awareness and that integrate with with the networking concepts that you recognize the network policies and all that kind of stuff we try to be a good citizen as well the being but being basin operated at that's another patent or so I just need and new part part important part is that we try to behave so even if we of virtualization of a translation conflicts with containers in certain fields but then we still try to behave nicely in the cluster so we also want decorative like you can declare parts of the state of points and we want when we follow this pattern mining inference in implementing the operative we so the last bullet point is the incidence ipods hand on 1 such an implementation detail but often does people to care about it so that's why named so yeah the ants are not called in in our implementation but the and live in the resource group of people what's the difference in what is the matter anyway so we have the ability to say that run and my comments about but doesn't mean it's thoughts of what you so it reports that common the only unit who need to swear off in in the case of of a specific look what Lotus's support all other concepts of build and ports so I will consider reports what we said is all right so communities track spots so we want to see especially tracks resource usage and it applies limits to what's so what we needed to make sure is that we really found that on the 1 hand communities is able to track the and the same thing way so that is what we place them in pots but you could still say well but why don't you for example implement Johnson communities and that is what we we get to to these conflicts again if we implemented the ends really inside connections then we have a problem that certain functionality is just not there in communities because it doesn't fit with those concepts for example if we think about like migration would use unite migration practice a few people it's very convenient because you don't have to bring down and calling customers they get will be back in 20 minutes but you migrated limited often works with a lot of time to optimize the process but it's not in the container for it you can implement that like I wonder how good it will be coming from but it's conception of something you have obtained from its status you how how the court yesterday immortal so they can go away and he just reinstantiated new 1 that is not something you want to do with the production the ends of legacy applications what to keep them around that's what we like so we want to have like migration but we knew that we cannot bring into communities because it clashes with the concept of containers and so that is why we say that hold the and logic is a topic and it's it's a layer of communities and it's leveraging the concept of communities and that's what this plot we see a part of the harbor for so you bring a plot and that and then that will replace the if we do migration which is creating new policy then we move the BM from 1 point to the other and communities Gilad afterward so the state of the again as we node not tied to a part which is good because the focal probabilities handles parts in containers but we can still deliver all functionality of conglomeration because we just see the as as resources constraints of resource contains 4 games if I'm getting too technical or computer please let me know Portugal you questions also no alright so at the beginning the that's good
and because we're getting I already to the end of the spectrum a full of units and we can get our hands on it would work so I'm confident those so has anybody tried
many you before all that good for all the people who did it and I would be recommended in cube so nice to try and if you want to get your hands of communities if he can drive it's really is like not just who will make you for being unique war inductive going to try and come so this demo I'm going to show the instance which again is based on a cube
around 5 using the Acadia which is just important was it's the oral and the notes and you can have different drivers importance to use the network output in the enlightened it can actually clone that them on all works to use well the this slide will publicly later on in case you really want to try
and so what I'm doing now is on that we will be quite small put what so what the demo doesn't it clones Cuba from up so we all public it's nothing can be pretty can be viewed as a complete um or so it's all public with all the issues there you're welcome to contribute welcome to try looking to find issues welcome to fix them and so what happened is with the that repository and use the my 1st so we tried all my so you we see and we've got the controller and we've got an RBM resources so sodium resources as our own DM definition DPI we saw before the control is related to the operator pattern so that watching for neuvième definitions and then acting upon them we've got the hand the which is sitting in said residing on every node making sure that the Vivian's getting scheduled to specific post that we know that we we launched in its constituent was and and all these components were now deployed on me you if you do it for the 1st time and take depending on the internet connections and undergoes images of pulled down and it's still early phase of the images are quite large and will get better over time so we look at what kind of put the running no that's to
bondage that and how we see that we've got a
few components running so the HA a
proxy if anybody wants proxy than live and of inward and with that as the mitotic because we I'll get to that in a minute with the deliberate spice proxy to give the graphical axles from out the cluster vertically I but control and I explained that the logic has been is actually that is the part which has set which is launched to provide the resource container for the and we've got the Burke manifest not yet which we want to get rid of all maybe that's and so these are the components and it's a stop me to and deployment so what can we do and look at the books and what into on and we can take a look at the liver and for example from and to just confirmed that we
really have been running at the and that's really nice if it works so how you see so what I did so we've got the word were launched at the end and that was sketch a specific calls with this plot for opening the Resource group for that year was launched they had solid and told delivered to spawn a B and based on the the the i which is also the case actually let's take a look at the Indian API 0 or that's because we've got our own typing communities that we can actually use the whole trip deal tools to to retrieve these commands and actually the goal of hubris country French citizens and we want to know what with Cuba the communities so you know today q c l is pretty much and what the quantities this so it has all the perks to work something but it doesn't support life migration for example so what would the communities guys we actually providing on put there to make sure that for example in cube CTL case we can add our own verbs and if you're tracking the progress and you might have seen that communities or keeps a now I think 1 7 1 8 is supporting binary problems on involves so you you we now have the ability to provide our or add onto cube CTL to then provide verbs which are relevant legislation so we can have over K budget CTL life quietly and beginning for example and that that you
so here say that it's really medium definition as as it was shown in this slide that is used to bring up the the among the most and if you look above it has all know has the feeling of communities even if we can express the all the requirements you can work with labels we can work with taken interations the so can we leverage the infrastructure what on the inputs so we started with that in the 1st slide said containers and literacy if you want to run them today in production we need separate infrastructure a sensible in Cuba it's not really used in the same likelihood of that that if you look here we see that some so this little this little section says that we
at that this could be the end but it's not using a persistent voicing from committees in this case but rather directly nights of the device on most of the time but what we work on and it actually works the which adopted is that you can use regular persistent volumes of communities instead of 4 4 plots you can use them for the atoms that certain betray constraints so much machines that don't support file systems this you need a robot device or device which is providing raw block it was like semantics and so so we work with communities and we actually now have included 1 of the 1st robot by support and we try to engage in a we try to really get it into committees on the long run that we have still support of time and you can use PCs with with the inspec so instead of referencing ask otherwise you'd be like you could use it for POPs so when the semantics of all pixels of all of the access modes would read would provide or or recognized what we currently working on because that's the 2nd part large infrastructure part is not working and that's actually more tricky than storage and we also working your collaborating with different parts to see that we get out of importance of the life of the infrastructure of communities so what does that mean we just want to connect in a simple way the ends to the same networks as communities is using more complicated than it sounds in all right so the demo was running
around we show the ends we showed that it was running if you follow the complete demo they can download binary tool and then you can do stuff like that of that you you know what do you know about and in clustering can't and
then you can do stuff like a little test the issue is that we like move to a new binary which is a three-stage to having that on the TTL which I did not installing it awakened directly access the consul in it even say that cube city of spies school in the
thank you spice into
then open the spice window all right and then in fact 2 slides
and and that's but yes it was about the the
demos of good but at the end because you never know what works so well yeah that's it from my side thank for much so
far other questions very adept at it at the top of your head high I think this was not the issue here is that that though the question was what is the problem of networking and the problem of networking is conceptual so so many problems of committees Indians are conceptual so what you get in a plot is not and you get an interface but that's this implicitly what about is getting core concept is that everybody is and I keep so community cystic taking that modest getting people but in the virtualization world you don't give not p to the mining you people might do it if you teach piece of land but effectively give it an interface so nature connectivity and so we need to say how we can so we've got actually is there a colleague of mine is implementing now layer 3 connectivity for BR so that can be playing ice stock in networking but the core problem is we need to address is we want to give the instability like in our classical data today to choose their own IT and just provide and related to networking when on yeah it's a week we could actually say and we don't want to do that we just want to keep the communities way manage it is for them to to to i ple this punishment but but we know that it takes time to legacy applications support of living with the fact that they know get an IP and taken dictated
anymore so but our 1st display nice because you get there and that will work with C 9 so who we are requesting you interfaces falls from for each snake which fortunately we assigned to be the that's going to work in progress 1 had shown next year work you can't do this anymore questions to the work of course the men and this be a and all and the mind of all also them on the part of the world and the rest of the of the of the of the you know kind so the question was why do we invest anyway if the integrand up with the question we should probably stop now and then and yeah I mean the assumption is that the ends don't die so quickly and we just even if it's quite some effort after all is not so much it's actually pretty nice to technical challenges of reduced from working over thank you for the things that I'm but we need the I think it's worth the effort because it he takes time and sometimes it takes more it we all know it often takes more time than you estimated at the beginning so we know it will take time and if we look at how many there are a lot of companies to get money from moving stuff from UNIX to Linux and Unix is pretty old so we know that transitions take take time and we want what the solution for this transition but we don't want to have you been stock of the infrastructure I think of the school would eventually stop supporting all of a sudden of the where no really goes up and just supporting pain as what you do with these words the 3 the serious not seeing any object in more that's what we say you still need to run Europeans but we also need to be up-to-date security updates and that kind of stuff for some States worth the effort to save we allowed you to move against the new infrastructure but keep your classically and running until you are and so eventually and there are reasons often again what we see that the Internet like for for really you want do testing on certain architectures see I'm such a big thing and does it works on the implication of all 4 for which contain a spot on lower levels just doesn't and this that is what we know that and will stick around also force isolation techniques we note that the isolation of containers because of the single kernel approach some say it's not as good as some say sufficient but regardless of the discussion there are regulations which required to have a strong which machine isolation eventually these regulations will seen because well but so far they have and in these cases to meet regulations of governmental populations you need to run the and so on and then the question so with regards to the government to make and then I said that was in that and the and I want you the the very lot and of the this and all flatpickers in front of them so does what was the question in terms flat characters is mostly back the all she's got too much time take look in fact we really need to give it some particular some of them so that there is no is a user-friendly offer users of container format will not contain formed the container runtime targeted for unprivileged users so you don't need from Texas from containers and the question was why can't we don't reduced classical pertains to run applications why should we have and I think it's after all to sing in uh technically they also you see what they also use namespaces we user namespaces that's why they can run on privilege which is symport characteristic but image wise the operation they might use different techniques on the low level but the core concept it's a container they should dependencies in a container so to me it's not so much different also something on this and that and I have to say about the the the line and the the and on top of that OK so the question was an idea that got on sorry and the question is why countries flat vector run densification it's difficult to run that they the Windows applications in fact if 1 doesn't support them and it's difficult to run that Power PC or for a common occasions in flat negative got 68 forms and that is where where again from the beginning so we said that initially the talk was what are the differences between the and things and it's like that so it really depends on the application if it's like the marks I would say alright you don't need to be in putting up you can use them if you run out or a column calendar application of the terminal you can use but if you use all even if can also changes the company and you can just that the spin-outs you can get it for the art and architecture you want and its little space that's no problem but if you really get to the details you want to have you have graphical application which requires a well and lately it depends on the requirements of the application if you need remote support what kind of a stroll isolation you know if you need a different color because architectural different features those those those questions going to the decision actually had diagram early on I would say it depends you can portal application directly connect to containers but often legacy applications to roll over years we cannot directly more than content as new transition period to adopt the implications of 1 hope I got a better any of rule you will now try to that anymore and I would like to see what have no very good if you've got any questions get us on our on is to put I there's Twitter you don't find useful because we all friendly well humans that is used here
and they will watch for the wall and fj