 Apologize in advance, you know before coming to the summit I got a call last Sunday So yesterday was now it seems as sunny, but yesterday was raining like like a crazy So I was walking to my hotel. I went when you know, I get very wet. So I'm losing my voice For tonight. This is it's a really tiny room I have a microphone and the session is only 15 minutes. So we should be safe But anyway, if you don't hear me, just raise your hand and I will try to repeat my my preso. So as I'm for unrodden us I'm a I'm a platform engineer working on the Clefondry project at Pivotal You know, I have been involved in open a stack for a bunch of years My first open a stack summit was back in 2011 in Santa Clara summit. I think it was the it was the design summit for Diablo Later back in 2012 I started to play with a club foundry project when it was open source I was an external contributor to the project and then I when I Get the hire but the company that is leading the Clefondry project And I have been always interested in the intersection of a club foundry and an open a stack Because I believe that these two projects are not an exclusive projects So they try to solve different use cases But at the end the core the root problem that they are trying to solve is exactly the same So if you look at what happens at different Technology technology shift during the last 30 40 years you will see that, you know back in the mainframe area in the mainframe era So in order to get computer resources, you need to work for a big enterprise That they have P money to invest in a mainframe and this is when you get computer resources later on the clean clean service era You know every one get a computer So it was easy to get that kind of computers that of a computer resources and lately if you go to the cloud era It's easy to get more cloud or a storage resources going to any kind of public offering like a Google like AWS HP whatever that what you can get on your own enterprise Also, if you look at the different Devices that are starting you will see that you know you have more power on your On your iPad or your phones on the different wearable devices Etc. What is usually happen is that the business people the now It's realizing that the the power that they can get from all of that devices Being able to gather all of different information being able to process that information get a lot of different Information for all devices is pretty huge and they want to liberalize that kind of information But the problem what is happening is usually the ID people cannot meet Despectations that the business people are asking that so what we what we filming belief is that We need a radically new approach if we need business people if we need it people Being able to wrap it in a way. We need a radically different new approach and You know that there are two different vectors of that one on the open size The developers what they need is to be able to innovate to fail fast To use new technologies to use new open source projects to try that once to fail to repeat To iterate etc. But also on the operation sites. We also need a way To provide that kind of resources to our developers to be able to scale that resources If the project wins well if the project gets a lot of you know success Then the operations need to scale the platform to be able to provide that kind of resources to the developers In order to do that You know if we look at the But a traditional legacy Systems you will see that in order to provide an application There are the different liars from the networking for the very metal servers To operation system the middleware the rune time the applications the data etc If you look at how most of the enterprises work every single liar Needs a different human interaction usually what happens is the application developers needs to raise a ticket That ticket is being sold but you know for any kind of a minister to that is going to provide a server This is you need another ticket to create a network for that server You need another ticker because there is some people that is going to install a middle-world liar inside that server etc etc So with with a cloud software and especially with with open a stock or whatever is you know, it's this is this this liar Get sold it gets over because you know usually what happens is you have a dashboard You have a second of self portal that you can go there and provision a server for you You know, I was When one hour ago, it was a Kenneth. Why at this same room? Yes, it was funny because he was in a room full of being where I mean people and he asked the question How many of you allow your developers to go to the being where I mean console and Spin up a new virtual machine and from a bunch of people only one being where I mean people Ask it. Yes, I allow my developers to that But usually what happens is now you have a cloud server like being where or whatever But I mean people doesn't allow developers to create the virtual machines. This is what usually happens No, so we with open a stack What we're trying to do is radically different So it's the user who has the power to spin out the virtual service if you want is the Po is He has the power To create a network an isolated network for for the project, etc But we also believe that this is not enough There are a lot of use cases that this is pretty powerful Just being able to let you developers to create a virtual machine. It's pretty useful But there are a lot of times out. There are a lot of use cases where this is not enough This is where we see the powerful of a platform as a services because sometimes application developers Doesn't need to deal With the middle-world liar liar or the room time liar They don't care about that The only one to deploy my application. That's what they need Nothing more than that. I'm not going to say that they are exclusive Paths because this is still pretty useful. There are a lot of times where Applications need to deploy some kind of code software a commercial of the show for a software So they don't care about the room time. They are going to deploy that kind of software, but the other use cases were not With it in the deploy just an application and this is why we think the platform as a service is going to help developers The developers they only need to deploy code more not nothing more than than that So how Cloud Foundry helps to realize that the vision so what we think is that For a developer it should be as easy as to target the cloud whatever cloud Just push the application In case that he needs to scale that application is just a simple common. See okay instead of one instance of my application Now I need 10 or I need a hundred of instance of my applications And if I need the service and you need the data services to run my application on my SQL Postgres or whatever it should be so simple to say I need the service Please bind my service to my application and that's all this is what what most of the developers Needs nothing more than that, but also if we look at the you know the previous version when we talked about the Operations vectors for the operation it should be also very easily and very fast to scale the platform if I have It's my enterprise. I have a you know a Cloud Foundry or open stack environment and it's on premise and I need to scale that Platform and say okay instead of having a 10 developers because I'm just better testing I need to deal with a hands developers or I need to support another kind of that the service It should be very easy to do that So what? Cloud Foundry provides so Cloud Foundry provides several largest one if that is a the automatic middleware configuration There's something that with Cloud Foundry with open source project. It's out of the box You will get a middleware to ruin your Ruby applications your Java applications your no GS application code Etc. But if there is a customer there is an enterprise that needs To support one of the middle words that is not in the Cloud Foundry project It's very easy for them just to create a you know that kind of middleware to package the applications and just deploy Applications so and this is something that it can be shared can be shared in the means that If you go to the g-hop if you look at for Cloud Foundry what we call build packs You will you will find that there are a lot of different build packs there that you can reuse And this is something that we believe it should be portable between different Clouds and not only Cloud Foundry the Cloud Foundry build pack systems is something that it's from Heroku if you are familiar with Heroku you can deploy your application to Heroku and then you can deploy also your application to Cloud Foundry without any kind of change And this is what we believe the cloud portability. It's pretty pretty important for most of the price Also You know lately there is a lot of booth around a docker So we have been using in Cloud Foundry. We have been using Linux containers for a while Since the one of the since we started the project we started using linux containers And what usually happens is when when you start running with with docker What happens if you want to run or just a docker container is pretty easy You know you can do a lot of the docker binary your docker cli just on your linux machine you can start a demon Create an image of the line one of the image and that's all and run it but what happens when you need when you Start dealing with lots of different containers and when I see lots It's about 1,000 containers how you are going to orchestrate all of the different containers In a single host in a multimodal host how you are going to get the locks for that in those containers So this is something that the platform like Cloud Foundry is solving for you Right now we are using linux containers We are not using docker in the next release of the Cloud Foundry project We are going also to super a docker image So it will be very easy to have you know a full cluster of hosts And it will be really easy to orchestrate your docker image from that batch of of clusters So what else you know when we talk about you know, it's about the service binding when usually the applications They always rely on a lot of services. They need that data service to run You know making an application that it's only code, you know, it makes no sense So what happens is Cloud Foundry provides an easy way to bind a service to an application It's so easy to say okay This is the marketplace of my data services that I'm providing to my applications And with just a command and say bind my service to my application and you are all set Also provide high availability on a scaling So one of the things that Cloud Foundry also provides is if for some reason one of your application dies Or an instance of your application dies Cloud Foundry is going to take care of that for you So it's going to restart automatically the application for you You have for example a hundred of instance for some reason one of the instance of that application is going to fail It's going to restart for you. But what more? What happens you have a cluster of different hosts and one of the hosts dies No, we talk about you know, and the panel was talking about the Femeral of the different virtual machines if you go to one of the cloud a public offerings to do for example, I do bless What happens in if I don't bless is going to restart your service This is something that happened, you know some some weeks ago, you know do the show look Okay, I'm going to restart one third or the full, you know all of the machines What is going to happen? What is going to happen is that in that case Cloud Foundry is going to take care for that for you What is going to happen is one of the hosts is going down Cloud Foundry is going to realize that if you set five instance of our application and there is only four because one was on the whole that has been restarted or has been killed then it's going to Create another instance of that application on a different host And this happens out on adequately also in terms of monitoring logs, etc What usually happens is in most of the times if you look at the previous slide when we talked about having Developers being in care of the middle world or the operation being in care of the middle world What happens when you need to do is a problem with your applications and you need to get the locks for applications What usually happens in more enterprises is that developers needs to open Needs to raise a ticket and say hey, I need to get the locks It should happen two things one is that I mean is going to get ECCH credentials to that developer You know in terms of security. This is not the best approach or the admin people is going to get the locks I is going to send the locks to the developer, but this is going to take a while So with with you know with platform micro foundry you get a fire hose of the locker of your application Also, if you are the mean people that are in care of the plafondry cluster You are going to also to get the fire hose of the different locks of the different applications So it's very easy if you have any problem with your applications to get the locks without having any kind of access to your virtual machine also, you know When we talk about having a big cluster of hosts and you Having a lot of different data services what usually happens is that you need some kind of network isolation Security to groups for example, you have a cluster where you're ruining your Developer applications and you have a data service that is for example is for production. You don't want that your Testing applications get access to the production environment So this is something that most enterprises happen. So the fundry also Provides you with network isolation and security groups and also provides you What what is that the role and policy management? No, so sometimes you In order to separate or isolate the different departments in your org and being able Just for a people on a single department to access the applications and not being able to see Applications of a different department with some kind of isolation. So club foundry also provides that kind of of role-based access and obviously, you know club foundry needs to run Anyway, and this is you know, we always believe that the best way to run a club foundry is on top of Open stack so club foundry is agnostic in terms of the infrastructure services It doesn't mean any kind, you know of access to an API an open-stake API or being where API etc There is another tool behind club foundry This is what we call Bosch that is able to deploy club foundry on whatever infrastructure And within also that the best way to deploy club foundry is on open stack and we also, you know very proud of You know sharing the the journey of club foundry with a lot of different companies and most of the companies is also Members of the open stack foundation. So we have created what we call the club foundry foundation This is something that we are still spinning up So three weeks ago we we share it with a community which are the bylaws of the other club foundry community Which is like islands which are the board and they are heavily inspired with the open-stack foundation And we're very proud that most of the company are also as IBM, Rackspace, HP etc Are also on the same journey of deploying club foundry on on open stack So, you know, it's that's the short talk is just about 15 minutes You know the best way if you want to try club foundry But this is a way to try club foundry if you want if you don't want to spin up your own infrastructure There is ways to just try one of the public offerings of club foundry It will be the people don't work services. It will be IBM Bluemix Anni-nines is another company that is working here in in Europe if you have some kind of, you know Constraints and in terms of that allows etc. And you only allow it to test your application or your data services in Europe You can try any nines and you have also the HP Alien platform also if you want to deploy it on your On your own premise And if you are going to try on open stack so the HP alien I think that last week know they released the nice installer on how to deploy The club foundry platform on top of HP alien ruling on premise And that's all you have any kind of questions. I'm open to that So so club foundry by itself doesn't provide an auto scaling system So if you want to scale your application, then you need to Instruct a fondry to scale your application application to a number of instance that you want Saying that there are some commercial offerings that provides an auto scaling and that kind of auto scaling It depends on you know different variables one could be CPU another one could be memory Right, right, right Right, so it's Or it so it is based on CPU is busy the memory or it's based on a schedule For example, you know, I know that my branches open from 9 a.m. To 5 p.m And only on weekdays, so I can program my instance of my applications Okay, I need 100 of instance of my application to read that period of time and Of that time we weekends, etc. Then I can scale down my application to just having one or no not all But this is also, you know, it's I was not going to talk because this is an opportune community It was but this is part of our commercial offering, right? Yeah We Wow, that's a complex, you know, it's a complex Scenario usually what we have is, you know in terms of the fondry what what we provide is some kind of application performance Monitoring and we do know we monitor different things we number the number of records that you know the world class is getting the number of Those kind of things work because there was a Finite number of nodes that they were able to do span ports off and then take every single transaction Right It would be very challenging because one of that, you know, one of the things that a foundry or open stack or whatever it's raised to You know to to enforce is that the familiarity of everything? So you should not care about in what host your original application, you know, and this is it's a really compelling use case You know, I don't have the right answer right now to to say, okay This is what you need to do Because it's not the way that we think that close applications should be the science for How to solve that? I don't know So you start doing complicated things like your own reservations or banking negotiations Right, right, right, right You know, I'm gonna say I don't I don't have the answer right now No, but it's a valid use case, right, right This is what you know, I usually take when I do the kind of presentation I usually say that you know a buffer on as a service is not for all kind of applications You know, you should be a very aware of what are what kind of system are you the sign it and based on that? Try to choose for it because sometimes it's not about platform as a service It's about having a cloud. Perhaps it's not the best use case for for that So right now we support from both Folsom to ice house Yeah, so there is a gap on the documentation I'm I'm aware of that so it's supported It's supported ice house because one of the our CI system is ruining on ice house So I'm pretty sure that it was a nice house It's you know, it's that the documentation is outdated and we need to date that documentation, right? You know, usually what happens if you look at the open study PI it has no change it From I will say that yeah, Folsom it has no change it but it's going to change with the bear also be three It's going to change but you know, it has no change that I always tend to say no the problem is not the open study PI The problem is how you have configured the open stack But there is so many ways to configure open stack. You can still use Noah network You can use neutron. It should change a lot. What the purpose are using you can use KBM You can use Chen and you know this well, this should be agnostic. That's not true If you get an IBM and attach a device to that VM and you are using KBM You will see that the device name. It's under the VDA if you go to a Chen a pervisor and you attach at this You will say that the device is deaf X Bda so there is a lot of things that that changes But you know in terms of the open study PI I will show you that you know from Folsom It's it's everything is supposed supported if the open study PI it's using the It's B2 when we change to be three. This is going to change Well, as far I can submit a pull request You should be perhaps tonight If I get to the room and I don't get I can submit a pull request today I don't think I'm going to the party. I don't feel all right, right, right and in a question Okay, thanks