 Hello All right, so let's let's get this thing started My name is Arturo Suarez. I I am part of the product strategy team at Canonical. I lead two products slash services there. I'm in charge of the training and I'm in charge of the fully managed cloud service called Bootstack for those who Might might have heard the term out there So between those two services are my two kids that keeps me busy 24-7 I'm gonna talk to you today about hard choices in life as as well as in cloud If you're here instead of Enjoying the Sun outside in beautiful Vancouver it means that you are Already making or you have already made One right choice, which is to join OpenStack to do OpenStack There's still one more choice to be made one decision to be taken, which is how you want to consume OpenStack and there are several drivers to that that According to the OpenStack Foundation survey If you look at this you're gonna find basically some words around efficiency Money even if it's misspelled there. We all understand what money is so pure cost This innovation this agility this flexibility It all can be split down into cost choice and time to market OpenStack brings as default brings as default Some of those Ubuntu OpenStack is the way in which you can you're gonna get all of them Let me dig into that pure OpenStack economics, right? We all understand This thing so both that price straight 15 dollars per server per day We can talk about what we mentioned in the in the first session of Mark Shadowworth About the price per VM afterwards, but basically a hundred nodes half a million dollars A list of thousand VMs, right? Let's go to the do-it-yourself way How much does an OpenStack engineer make? Well according to indeed.com in the US Average salary of an OpenStack engineer and in twenty eight thousand dollars You're gonna need four to six of these operators that depends on how often they get sick How often they want to you know take a day off, but basically four to six operators that concept to Half a million dollars the best scenario it would be something like half a million dollars, so it's very Comparable all right put stack makes sense for 80% of those clouds. So there are very few clouds That have actually more than 100 nodes Okay, just pure economics But you might say hey, well, I want to have my team I want to hire my team right and this is a picture from the previous OpenStack design summit in Paris we had this window on top so we could take this picture and guess what they are also hiring all of us are hiring down there Right, so you've been competing For talent with all the products all the vendors all the companies around OpenStack and that makes your life easier I mean you can think as well. Well, but I have a talented team already. Oh, sorry There's there's this thing about hiring the team and then training the team, right? You might know that we have these sessions like every six months with a lot of new features all rooms Are are full right? It is hard to keep them trained It is hard to have all the latest versions of OpenStack ready and all that innovation Ready and available to your customers, right? You might think well, I have a talented team This is LeBron James four times MVP of the NBA two times and NBA champion Juju deploy in OpenStack. He's happy. He made it struggled a little bit with the Neutron high availability But he made it right The city of Cleveland and the Cleveland Cavaliers might want him to still be playing he plays tonight And winning basketball games instead so you get my point, right? You might have a right a good team. You might not want to allocate it to something that might not be core to your business So we cover the cost We're gonna cover the choice the second of those elements, right? We have something called the OpenStack interoperability lab where we build 3000 clouds a month combination of permutation of all the vendors all the software all the different products or components of OpenStack to make sure that those Interoperate that those are fully integrated right that choice These are some of the partners we have there and that choice is then translated into options in the hypervisor We obviously 80 something percent our KVM that will change with a new introduced this week Lex D Lighter visor where you get 14 times the density and Hyper V were able to do to deliver Open second hyper V now in terms of storage you got Swift you got Seth you got a lot of other vendors coming along that have also joined our open second to the ability lab and in terms of networking This is a bigger a bigger The component that's getting most of the attention We can deliver clouds of course with the default open V switch, but also with juniper contrail No, ah from market elucent Meta switch, which is a very interesting as the end Mido Kura, of course Plum grids Cisco and I think last time I checked there were 14 of these as the end vendors in our OpenStack interoperability lab Important thing to keep you competitive. It's time to market Which we translate into time to production OpenStack, right? So let's go through the process you hire we've been doing that raw show recently some of you We met in different cities of the world from Seattle, Brazil, Mexico, Toronto Talk your Singapore London. I can't keep track of all of them, right constantly We've been asking people how long does it take to build an OpenStack team and Regular is something between four and eight months, right and then of course you train them and then you deploy and Then you got your open stack cloud in production, right with a managed service with bootstack We deploy in a couple of weeks you get an OpenStack in production and then you can hire the team and you can train The team right so we turn months into an OpenStack production environment into days Bootstack stands for build operate and optionally transfer your cloud again We're going to introduce one of the concept that people worry about and one of the drivers that Define how you choose to operate your cloud. So let me dig a little bit into that how we build clouds, right? We have our eyes in many clouds In fact, we are the ones that have eyes in most of the clouds, right? 55% of production environments are based on Ubuntu OpenStack. So we've learned we've done this for four years we have Distilled our experience into a reference architecture into a way of deploying OpenStack that accommodates not only the Workloads you're thinking of right now, but the work was you're not thinking of right now, right now It's a our balanced reference architecture a balanced way to deploy OpenStack We'll accommodate those workloads. We start to see and this is another again at a sign of maturity of OpenStack We start to see some clouds failing to be upgraded and failing to scale Right those initial clouds that were built two years ago when year ago are now starting to break because they cannot scale more and OpenStack should be scalable or hyperscalable We operate it completely Let all levels of support you're gonna get this very same level of support that all our main customers there Get right. So these are the guys who set our standards High standards, right? And this is interesting. This is very unique feature We transfer the cloud back to you. So one of the drivers if you remember was lock-in, right? We're not about lock-in if you want to transfer your cloud if you want to have your team and operate it We don't really want I mean we we can operate your cloud. We know how to do it But ideally when you have a team to operate your cloud We will step back and continue to deliver all the innovation of OpenStack and all the innovation of the vendors around OpenStack To make your cloud even more successful, right? So we hand over the key and you start driving we have again my other little product here the training So we will train your team. We will make sure that that transition is smooth and you Well start operating it from that point over whenever you're comfortable. We're happy to transfer the keys, right to illustrate this. Let me introduce one of our Customers so my media CTO might tell you and he's gonna tell you about what they do at my media and how they Make the most out of their Managed private cloud. Thank you Thank you very much, okay, maybe I'll just just this one little slide Okay, so okay, so Just to introduce what my media is so my name is Michael Yoon and I am the CTO of this company. It's called my media So at my media, we are a company. That's enabling Customers enabling end users to be able to rediscover their digital memories and we bring the perspective here at OpenStack of a consumer of a and we're providing a production environment for a consumer cloud and Utilizing the the OpenStack platform So we're a little bit different from many of the other consumers of OpenStack because we're not providing a set of cloud services to tenants We are really our own tenant. We're the own customer. We have a large application And we are you know building OpenStack so that we can use it internally for this application We also are one of these you know pretty exciting startups that are sort of on the cusp of breaking out into a very very large year We've got major major scale requirements So we have expected storage capacity by the end of or within the next 12 months of about a zettabyte of data So we're really looking for a large amount of capacity and expansion Driven by both organic growth, but then also some major partners some global partners through the through the course of the next year And these partners are our sort of device-based OEMs ODMs as well as telecom carers and retail partners So what this means is we have a need for obviously a large amount of infrastructure And we are very much interested in choosing the right set of partners So why do we choose canonical? So there are many different reasons, but three of them that I'll highlight here Certainly the first and foremost is the model of engagement. So No, no joke. I met Arturo yesterday for the first time in person and the words that he you know Sort of illustrated there are exactly one of the main reasons why we chose canonical So the model of engagement, which is the build operate and transfer model is exactly the model It is that we've been looking for So we want to focus first and foremost on our application and at the same time we have the needs to be able to build out this large set of infrastructure and The build operate and transfer model is exactly the kind of thing that fits our business needs at this particular time The second thing is the breadth of canonical's expertise, right? So not only are we utilizing Ubuntu, but also There are many synergies not only because you know canonical doesn't provide just boot stack But they've got a range of things a full set of services including partners You know relationship with vendors and a number of things along those lines and then finally the Embracement of the culture that canonical is embracing which is really about agility and customization Really fits what it is that we need right so because we are our own customer and because we are not providing sort of a set of things Services for tenants an example of the types of things that we are perhaps a little bit non-standard is that we're looking to be able to deploy Our open-stack environment on a set of flat VLANs right as opposed to the traditional sort of VX LANs things of this nature And that's not something that most reference architectures or most sort of you know You know, we know the best way of doing it and here's the way to do it But canonical really embraces this approach of understanding what our needs are why and then working with us closely So those kinds of things fit really really well with an agile sort of Philosophy and the startup needs and the needs of a startup company like ourselves So those are the three main reasons why we've chosen canonical as a partner and very happy to have done so We are also telling a bit more of our story So if you're at all interested in the story of my media and the other sort of you know things that we're working through We have a talk today at 240 in room 204 Thank you. All right. Thank you Michael So one of the of the second derivatives All right, so one of the of the of the immediate consequences of this ability to deliver fast to create Flexible clouds to create balance clouds has to do with all these choices all these vendors, right? So we are keen to Deploy several times for us It doesn't take too long to deploy several times pretty much the same cloud the same base Infrastructure the same hardware the same underlying infrastructure with different sdn vendors So you can have an sdn evaluation tool Pretty much in no time once we have deployed once with open vSwitch. We can redeploy that with open control with some other race It also applies to other Components of the stack we have a little bit of help here You probably have heard about juju our universal modeling tool That allows to deploy different services in different clouds in different substrate one of those substrates is of course open stack This is a juju charm a juju bundle of Open stack with open control, right? That's our the components of open control from juniper This is pretty much the same bundle, but we are leveraging here Calico from meta switch a very interesting of real for sdn Same one with noise from our catalusant Okay same one Same open stacks in bundle with open daylight. This is the way Cisco would like to Deploy their their class the way they actually deployed their clouds to illustrate that again, I'm Happy to invite home stage Garima can see from pure one another of our of our customer Thank you very much So my job appeal one is I'm a senior architect and principal responsibility at the moment is proof of concept and evaluation of cloud platforms now we are a Very large hosting company We have a very large network and it's very important to us when we're deploying open stack one of our serious concerns is we have something which integrates well with our network something which leverages all our network services The answer to which stn does that best is Something nobody can tell you it's very unique to each company is our feeling about it certainly and The end result of that is we need to try a lot of them and despite being a relatively large Organization for a hosting company certainly we have a lot of resources in engineering and architecture It's it's a lot of work to rebuild an environment repeatedly to test sdn's and It's not always an effective use of our engineering time Given that it's using the bootstack service has helped us speed up our Testing and evaluation process there quite substantially It's been a particularly Smooth process due to the level of flexibility and customizability about it As has been alluded to already the fact that there are multiple options for really every part of it Whether it's hypervised or a storage, but particularly in the network space that's Canonica willing to build really to our needs is hugely valuable compared to some other partners who Michael alluded to a very much in the model off. This is how we do it. If you don't like that, you need to go elsewhere So that's been Really quite a smooth process for us we have also been impressed by the work Canonica will put in around juju and Turning our sdn options into charms for us so that we can See what the deployment will be like in production. That's obviously it's a Big deal that we can see how smooth the production deployment off these systems would be And it gives us a lot of confidence when we're testing the systems Yeah, I think that's the overview of what we're doing with it and Why we see it as having a lot of value for us Okay, thanks. Thanks, Gary Again with options we also have options how to consume open stack This different different models, right? So we have data center hardware so bundles of the of different different components of the of the stack, right? So if you want to have a hosted cloud with Boots stack, there's a white paper Recently released. We are able to deploy that in a very reasonable time at soft layer Which basically it helps you get a Basically, no no cupex cloud up and running With a minimum small minimum commitment One of the companies in the hardware industry that actually understood this flexibility and a reference architecture Perfectly is quanta QCT. So they're able to provide a bundle boot stack on their quanta hardware They named it quick stack And you can purchase that the full Cloud in a box from them We have data foundry. They will give you the whole thing, right? Their hardware their colocation centers their data centers in Austin and Houston, Texas And we have our partners so agility networks in the room Our friends from alliance IT in the US that can help you Not only deliver this service, but also Services on top of that cloud some migration services, etc And of course our friends of Fairbanks in Europe in the Netherlands. They're bringing this Boot stack to the to their European customers We have some others in different parts of the world as well And of course we do you we're able to deploy things in what we call above the cloud, right? So if you are a hosting company, you can bundle a Tommy on top of that bootstack and you have a fully functional hosting cloud business Sweet We have charms for IBM software. We have charms for Microsoft software. We have charms for IMS and and if be Things we have charms for Different monitoring tools that would go above the cloud all of those can be bundled together in Services by our our partners So I believe I Am standing between you on lunch if you have any any questions at this time Or you want to go straight to To the nice meal, please go ahead. Happy to take them. All right. Thank you very much