 Excellent. So good afternoon everyone. So my name is Regis-Eleg. I run the software engineering group at CloudWatts. So we're a French-based cloud service provider and we're gonna talk to you this afternoon About our journey in order to become a good open-source contributor and in order to do that I'm gonna pre-co-present with Mr. Loïc. I'm gonna let introduce himself. Yeah I am Loïc Descharis and my day job is to be a self-developer. So I write code all day long and Something else I've done is being old. I've been a free software advocate and contributor for 30 years So that's what I brought to CloudWatts Okay, so first a few words about CloudWatts So what we are is a sovereign cloud provider. So created to deliver competitive. Yes services In basically in data centers that the NSA cannot investigate So which was quite important for France and Europe. So we're bringing those services to French and European companies In order in in terms of financing So we're actually a joint venture between two industrial partners So one of them is orange the telco The other one is TALES so the defense group and we're also supported by the French Government that has invested significantly in us and the differentiating factors of our cloud is around security privacy and resiliency in order to be compliant with the French and European regulations and One of the things that also makes us standing a bit apart is that we were Started with a free software DNA from day one meaning that all of the stacks that we've deployed in order to run our cloud Is based on imposter software. So of course open stack being the main part but also for instance delivering safe and these days open control and Today we're one of the largest open stack deployment in the EU. So we've launched last Last September a storage solution based on based on Swift, which we've opened up both as Swift APIs and through S3 APIs We've also put a SAS solution on top of that which is giving box functionality and at the moment We're running our compute services based on Nova and the other open stack components in beta mode And that beta mode is going to be converted into a full commercial offer at the end of at the end of June Where we're basically going to do is turn our billing services on and start measuring SLAs But at the moment we're fully in production and deployed Okay, and sorry and and one of the what the one of the key things and it's quick question that we had to ask ourselves was as Public cloud providers we could have chosen to be only a user and operator of open stack But we actually did decide to actively contribute Setting up a development capability and there were two kinds of reasons that we did that for so the first ones were necessity reasons We wanted to master the solution were deployed on and we found that the best way to do that was to actually Build a team that would be contributing into the various cloud components that we will be leveraging The other part is that it is very important for us to drive our Roadmap and product roadmap especially especially around the cloud functionality and if we wanted to have any influence of that We had to be able to talk to the development community And the best way to do that is to set up a group that actually contributes and is able to do that And the last part and that's a big part of our mission The reason we have founded is that there's sovereignty reason behind that meaning that we're also there to actively build Cloud computing skills and cloud computing development skills In our territory and in order to do that leveraging and and contributing into an open source project was the best way to do this What we actually get out of that were a few bonuses that Were quite Interesting for me since I was coming from a very much a proprietary software background And which was that the first one is that that gave us access to world-class technical coaching meaning that through the Contribution process and the review process that actually gives very strong Advice to our developers and really enable them to develop much faster that it would have been able to do in In a proprietary environment The other part is that this gave us access to leading edge Engineering practices and tools meaning that today what is best in terms of continuous integration and software engineering Process is actually coming in open second. I mean when you look at the Let's say advanced features that are within the CI the scale of it Etc. It would have been very hard for us and even impossible to actually been able to rebuild that ourselves The last another part we saw Is that it has really enabled us to boost the attractiveness we had for the right type of developers meaning developers that are passionate about technology and around cloud computing and As soon what we notice is that as soon as you say that you're gonna Be actively promoting open-source contribution you suddenly get access to much more interesting Recruitment pool that you would be able to do in your typical. Let's say localization and money conditions And the last one is that it also can provide you a very interesting leverage ability on some developments that she's decided to do in the sense that if you have good ideas and Since open-source is very much my retro currency base those ID could actually be and the answer Those ideas could actually end up being developed or could evolve by other parties than yourself So you actually have a much larger reach that's what you would only have with you on So I'm gonna now let Loic go a bit in more detail in some of those topics It's more it's more examples because I'm a practical guy So you had theory strategic decisions and so on now I do things and I will give you a few examples of how free software benefited in a concrete way Cloud what my dead job is to develop on safe. These are cephalopods and One of my mission was to reduce the number of discs you need to store you user data A cloud what is aiming at buying tens of millions of euros worth of discs? How do you reduce that actually there there is a technique that is fairly well known which is called original code So it's kind of red five which everybody knows but expanded to the cloud That sounds fairly easy to do It turns out to be very complicated to be something that requires a level of mastery of the technical solution That was beyond me at the time and even now to be frank Even though it's complete. So how did we go about that? well instead of going to a vendor who would sell this capability for a lot of money and lock us in instead of Buying a lot of very talented Self-developers stealing them from ink tank What I did is I started the work in the open as a free software developer and I faced quickly People who were much more competent than I am Sage Vale is most definitely a genius, but he's not alone. He attracted Samuel Just who is the lead of the core development and who is incredibly talented. So I got to Tackle a problem that require a level of mastery that I didn't have and I met halfway people who actually had the skills The two go hand to hand it couldn't have happened if I didn't make the first step and When when I started the work, they were kind enough not to point my shortcomings and let me go Have the way somehow it may be convinced the marketing people That it was a good thing for ink tank to write these goal in the ink tank road map And that's how six months later. I show I saw that ink tank was planning to Implement a ratio coding. So I was no longer alone with cloud what on this goal But the community emerged and in the end now two weeks ago. It was released and we have it Can you yep? That's yeah, I'm very happy about that very lucky to So another example completely different at the very beginning of cloud what The question was that the mission of cloud what is to create a sovereign cloud in order to do that you have many possible ways and Although free software is fairly well known in France most of the Management Does not know how to leverage it properly So it could have been the choice of the management to go for a proprietary solution in order to build a sovereign cloud Instead because we discussed first we highlighted that Through free software and open stack was a very a great example of that they could Have a technical solution that would be impossible to develop even with a lot of money and So they would have the power of this solution without the ties of proprietary Vendors That was one piece of the puzzle But now if you have just the technical thing and you do not have engineers to do changes you keep looking for people to Modify the solution to fit it to your needs And in the end you are no longer Independent you are no longer able to do something without asking permission to someone so that The two the combination of the two was a decision that is for me on the managerial point of view something that is The best possible use of the free software concept a Practical example of that would be next slide What happened in neutron? so one of the Stakeholders of cloud what is orange orange is a very large telecom company Who has an experience in networking that predates cloud even so there was a big pool of talent there Not specially skilled in free software actually very few although they had Large interest in open stack It was difficult for them to contribute because it was a big company and they they didn't see the advantage Now there was cloud what who chose this Technical solution they chose to invest in engineers to keep their independence So there was scale and a new culture and the two were mixed Edouard who is sitting here actually migrated from one company to the other and some Orange employees Started to look forward to contribute to free software We're today at the Juno summit and There are cloud what employees contributed to the code and orange employees contributing to the code so these examples shows how it Created something new the culture of free software appeared in both companies second to last example I'm too long. Okay. I will speed it up I've been worried all my life as a developer about the process of development it's always crappy and The reason why it's crappy is because nobody want to care about it when you're in a company You have ten developers who wants to define the process nobody Well, it turns out that open stack is so big that It's very we are rewarding to care for the process because you have a very large audience and The outcome is something that looks very familiar to all the developers here And also very scary when it's the first time you look at it now for each box There is a reason why it is necessary and it's being refined by dozens of people It's being practiced by hundreds and thousands within cloud what What happens is that this is the process that is used for development Not for decision-making you you have this agile process that is different So these development process it shows the flow that works for the community and why not for cloud what and it's not something that is All Encompassing you're not forced to kick everything out. You can use something Next to it. So that's another way free soft the free software culture penetrated cloud what and last It's not enough to be aware that free software is great. It's a cultural leap When you recruit developers in companies Most of the time they have worked for a decade if they are good in the proprietary world and When it comes to contributing to free software project, it's very different the interaction with the colleagues are not the same The social conventions are not the same and as I mentioned the technical tools are not the same So from day one we established a course that is available to all cloud what employees By which they can just Speed up the learning curve to become good open-stack contributors Actually this course goes even beyond cloud what since past weekend we spent a full weekend with 25 new open-stack contributors doing this exact same course and it was not cloud what employees dry HP etc So these are the examples I have for how it applies in cloud and So this is where we are today. So today. We have of cloud what a nine people team That is actually whose sole job is to do contribution So just to give you an idea. This represents approximately 25% of our overall software engineering capacity the rest of it being Spent into developing all of the commercialization aspects for our for our public cloud So one thing that that is important to understand there is that in the way the team contributes all of the user stories that our people are working on are always things that are making our roadmap process meaning 100% of the Actual features that our people are working on are actually stuff that matters for the for the clodra cloud watch roadmap, which is actually the only way you can get past through in terms of Justification and budgeting and so on However, what we're also doing is that we want to secure that all of our people are good open source Citizens so to speak so we're making sure that we freeing up approximately 20% of their time in order for them to do To have the tend to do reviews to have the time to do bug fixes and to have the time to develop the ties with the rest of the Community so that when we have something that matters they actually can be influential So today that makes us the 18th contributor on the ice house release We've had 25 proposed blueprint and five completed ones and on on sef So mostly thanks to to Loic and his crew. We're the fourth contributor. That's however one point of Think they're around those kind of vanity metrics that they're extremely important for Let's say service companies because that's what they build their let's say credibility on for us It's much more of an objective measure of how much we're doing But it's not something we can really use from a marketing perspective meaning haunt your selling cloud services People don't really care how much you actually Contribute yourself, but it's interesting for us to understand where we actually stand. So Yeah, and There's another important aspect. I wanted to cover which is that's no, that's actually Readable over there. That's great. It's how it actually Influenced your product development Your product development process and as soon as you have a significant capability that is invested into the open-source project You always have three options all of them We let's say advantage and and and cons So you can go upstream where the feature will be submitted and get through the acceptance process with all of the let's say Things that that this implies you can do downstream meaning you take the latest and greatest version of the code And you do the modifications you need of that or you can decide to implement it as an external component Where it's going to be integrated via the project API's and actually there's pros and cons with all of that And what's really important is to really understand what are the consequences or the choice you're making? So if you decide to go upstream, so there's a lot of advantages to that Which is that your feature will be reviewed improved and validated It's going to be maintained in the product by construct Which means that you have not created any technical depth for your for your services And it can you can also sometimes benefit from the leverage effect meaning if it's a good idea You may have other contributors that are going to help you implementing it There's however a couple of things that you lose as well when you decide to do that If that idea was unique and could have made a difference for you in the marketplace You've just lost it because you've given it to everyone else So you need to think carefully about what that means And the other implication is that you do not master anymore the lifecycle of the feature meaning you cannot control When it is going to be released exactly So if you decide to do one of the two other scenarios, so you decide to go downstream So you keep the differentiating advantage and you also can own the life cycle meaning you can go as fast as you want But you do that with a very significant technical depth that you incur and actually the let's say the fastest The open source project is going the most velocity it has The biggest that technical depth is going to be and actually increase Because you'll need to reapply it with a recurring cost every time there's a new release And if you decide to go at an external component, which may sound better as well You'll also be very exposed to potential technical depth not so much this time related to the speed at which the open source project is evolving But you're going to be exposed to that depending on how the APIs are stable. So you all need to care to be careful about that In our in our cases, we've actually implemented Depending on the cases each of those if trials of scenarios So we try to go upstream as much as we can and for instance We had significant contributions into the horizon dashboard because it was not necessarily suitable to operate a public cloud When we took it on and we through very targeted contribution We've been able to increase that in the last open cycle really cycles We've also done some critical things in both neutron and open control in order to Be in a position that we would have full API coverage from neutron leveraging open control, which is our target as the end So there's a couple of examples where we've been able to do that In some cases we've had to implement downstream changes. So typical case security patch But we try to limit that and always put some kind of time stand on it after which we cannot go meaning We also decide to do that with a clear and dainty mind after which we move back to the regular extreme flow And the last part as external components is that for certain component that we considered differentiating Such as some components of our security architecture We've decided to implement them as external external components integrated through APIs Okay, and in order in order to finish I also wanted to share with you some of the hardest question that we had to face We've regard to both investors and stakeholders When we decide to go to this to this way so the first and one of the most common question Is actually your tiny fish meaning that do you really expect to be able to influence anything? When the big guys and a much larger company than you are actually gets involved into that and and the truth is that at least our experience is that if you Take the time to bring that culture into your company and train a small number of people so that they can be legitimate We've actually been able to pass some of the key features of our roadmap to the regular open source project So actually this part I size we've been able to get some successes there Another very common question is the when whatever big company name you put in there Contributes it's actually a rounding and or on the PNL meaning that for a company like us when we're putting nine developers on open source contribution complemented with 14 to 20 external service contractors from companies like in advance that's actually very significant with regard to our development budget and Is that something you can really afford and for us? The reason we do that is really we don't feel that we have a choice meaning that Because we've chosen to base ourselves on open stack If we want to have any control on our roadmap We need to be contributors because if we want to have any control of which features we're going to be able to bring to our clients Another one is The challenge you have I mean as soon as you start lever it's it's a bit like the is around company culture versus the loyalty to the open source project and It's really the fact that You get let's say a massive boost in attractiveness as soon as you say that you contribute But you're also exposing your people a lot more because people will see that they contribute They'll start raising a stack and then very quickly they'll start to be reached out by all kinds of companies because they're interested in their skills and And the thing is that also what typically they're very attached to the To the project that they're contributing to so it actually has sometimes that could have more importance Then you run company culture. So the answer to that I mean is that we really trying to see that as a bit of a challenge for us to try to build an Environment that is attractive enough so that people stay and to build an image of where we would like to bring cloud What that is that makes them stay but it we've really tried to see that as a bit of a positive challenge In order to make a place that is attractive rather than trying to hide our people in some closets so that nobody ever gets to talk to them And the last one I mentioned there, but I'm sure maybe some of you actually got Got it from their management before coming here is that well, that's a lot of people you're flying to the summit So in our case we have so from that nine people team We have some of them that are going to participate in the design session So we kept a couple at home in order to check our production systems The but actually the the best answer we we actually Got to that question is that the choice was made before meaning Once you've made the choice to set up a contribution team It is essential that those people are actually active members in the communities and the best way to actually Talk to their colleagues and be legitimate within that environment is to leverage the opportunities They have to have some face-to-face meetings in the design sessions So, yeah So this was it So don't hesitate to visit us. So we have a small booth in here 31 So please come and talk to us. Don't hesitate as well to talk to our engineer in the design sessions We'll be very happy to provide you with beta test accounts So you can test a bit what our class look like and we'll be really happy Loic and I to answer some question if you had any many questions Yes, yes Who didn't we didn't exactly but and and and the answer to that as well is that those are people We would never have in the first place if we didn't contribute. So it's it's really the Absolutely, yeah, and and it's also I mean what what what I see as well Yeah, and people also live for the good reasons meaning they live because they have something much more exciting to do somewhere else and We're kind of fine with that if you see what I mean so All the questions Well, it's yeah, so, I mean those are questions we had so basically what what did I mean what we had to do was to think long and hard because Some ideas are quite clear at the beginning and then as you progress and I mean organization always evolve and change and some strong visions and actually change and you need to justify yourself And you need to have strong reasoning so and actually but the the the more Smart the questions get the more you need to think about it and build something meaningful basically Anything else, okay. Well, thank you very much