 A couple of weeks ago, I was chatting with Colin and others and I realized that we were talking about open source sustainability in general as a topic for the conference and a theme we wanted to touch on, given how hop-on of an issue it's been over the last year or so. And as I was thinking about that, we realized it's been over a decade since MySQL was purchased by Sun just under a decade one since Sun was purchased by Oracle. And I started to think I was talking to Peter about the Percona sponsorship and about their presence here and talking to the MariaDB folks about all of their presentations. And I realized we actually have quite a few leaders from the MySQL diaspora, as I was calling it here. And that this is an interesting one to me because it's a very vibrant community, despite the fact that it all sort of branched in a weird family tree about a decade ago. And yet, when I decided to write some code to query databases, it mostly just works across all of the different database distributions. And when I start to think about what databases are, the default around a lot what we use in Linux, again, MySQL or MySQL derivative or branch of some sort. And I thought this is weird. I don't think I've seen any other open source communities that have done it in this way. And I'd love to get the experts in the room to talk about it. And so Colin Charles was kind enough to help organize this and put it together. So we've got folks from Percona, from Facebook, Microsoft, Alibaba. And of course, Maria Dibi. And so with that, I'm going to leave you in the capable hands of Charles to start an interesting conversation about how we got here and where we go next. Thank you, Ilan. Good morning, everybody. Thank you, Ilan, the scale team for having us here. Many came from afar, even on this panel. And it's probably not a stretch to say that on this stage, there's probably a combined 100 years' worth of database experience from the people that are here. Some of you are still streaming in, thanks to probably daylight savings time. You've got one less hour worth of sleep. This used to be a problem with database replication, possibly a long time ago, though I guess now we fix that all since we all run with UTC time, right? All right. So Oracle apologized that they could not be here as stewards of MySQL, largely because their executives don't work on weekends. So without further ado, we start with the first panelist, Peter Zaitsev, CEO and co-founder of Percona, formerly of MySQL AB, and author of this wonderful tome called High Performance MySQL Third Edition. To the right of me, we also have Yoshinori Matsunobu, production engineer and MySQL tech lead at Facebook. He is the man responsible for leading MyRox development, and he's going to tell you a little bit more about what MyRox has done for Facebook. They migrate, they use the database, message, and so forth. Also, formerly from MySQL AB, also made a wonderful tool called MHA for high availability, still in use today. And it's his first time at scale, so give him a warm scale welcome. We also have Sunil on my left. He's the principal group program manager at the SEO data team responsible for MySQL, MariaDB server, and Postgres. They recently also acquired Citus data, showing the commitment to the Postgres community, and it is also his first time at scale, so give him a warm welcome. Thank you. Furthermore, we have Jimmy Yang, team lead for PolaDB, storage engine at Alibaba Cloud, formerly also an NDB architect at Oracle, and it is also his first time at scale. Someone should convince him to tweet as well. And all the way to my left, we have Vicente Keobaro, software developer team lead at the MariaDB Foundation. He got into hacking on MariaDB server via Google Summer of Code, and he has graduated to being the program administrator and mentor for the Google Summer of Code program at MariaDB. It is also his first time at scale. Okay, oh yeah, and then there's me been around for a while, and I've had the pleasure of working at all the server manufacturers, so yeah, next slide please. So that's the one thing that I'll give you from Oracle. So several years ago at Open World Keynote, Larry Allison came late to his own keynote, and then his clicker would not work, and then he would keep on saying things like next slide please. It's a wonderful video, you should go check it out on YouTube. So just to get things in perspective, MySQL's been around since May 1995. Anybody used it since then? 95, wow, excellent. So Oracle actually acquired InnoBase in October 2005. InnoBase makers of the InnoDB storage engine. It's the engine that probably all of you use. How many here don't use an engine that's InnoDB? Besides maybe Yoshi. Okay. MySQL also did acquire Sun for a billion dollars in January 2008, and Prokona server's been around the first ever branch since November of 2008. Oracle then acquired Sun, and it took a very long time because there were fears that MySQL would be killed at Oracle. And the MariaDB server actually came around in February of 2010. So, next slide please. Yeah, so is MySQL tired? And that's what I have as a question for the panelists because Prokona has also dabbled with MongoDB and most recently Sun offering Postgres. Alibaba is making PolarDB. Microsoft decided to acquire Postgres company, not a MySQL or MariaDB company. MariaDB itself doesn't contribute or more accurately cannot contribute upstream any longer. And this particular slide that some of you may see, not on the video stream, is when in 2011, Michael Stonebreaker himself said, he famously said that Facebook is operating a huge complex MySQL implementation equivalent to a fate worse than death. And the only way out of it is to bite the bullet and rewrite everything. So naturally, Facebook doubled down on MySQL, made T-shirts and mocked him. So, what do you guys think? Is MySQL tired, Peter? Well, I think things have been different, right? So, then I started to be involved with MySQL, started Prokona back in 2006. The MySQL was practically the only game in town, the only relational database out there. It was kind of Postgres, but that was much more niche database. And if you look at the companies found at that time, Facebook or Twitter, Yahoo, right? All there MySQL. Now, more than a decade later, we have much more choices out there. There are many good relational databases, MySQL, Postgres, MariaDB, and also many different kinds of non-relational databases, even many kinds of databases, right? We have graph database or database focusing on the time series and many kinds of no SQL database, right? And I think while MySQL is not as dominating that as before because there are many wonderful choices, I think it is still going strong. And I think those choices are wonderful for all of you. And they're also wonderful for MySQL community because competition brings up best in developers and in companies. Yoshi? Yeah, I think MySQL current state is better than I expected 10 years ago. So 10 years ago, I worked at MySQL as a consultant for MySQL software. Then lots of people said NoSQL would take over MySQL database, so MySQL had no future. 10 years ago, many people said that, but now MySQL is still everywhere. So even at Facebook, so there was a discussion that HBase was a superior database, so let's duplicate MySQL and put all engineering efforts to HBase. But actually, after a couple of years later, the opposite thing happened. So now the largest HBase back in Facebook was we migrated from our Facebook messengers, HBase to MySQL called MyDogs. So actually, it's pretty interesting. So many people, what many people have predicted didn't happen after 10 years. So more workloads for MySQL at Facebook. Yeah, actually. And what about from a cloud standpoint, Sunil? That's what I was gonna say as well. I think being such an old, relational open source database, MySQL has garnered an ecosystem that is second to none. And people only think about how do we replicate that? It's so pervasive and the path that MySQL took early on trying to be a pervasive, like an embedded database I think played out very, very well for them. And then slowly graduating to more and more B2C or business to consumer type of applications. And I believe the next, I think the accelerant to MySQL has been the cloud. And it's in fact not just MySQL, it's been for all open source relational databases. Now, making it much more simpler and easier to use and then compound that with other cloud benefits around security, performance, auto tuning and so on has kind of given a new life for these open source relational databases. And from our perspective as we operate a very large cloud, we see MySQL being used in many diverse type of applications and continues to be one of the favorite database engines. So from our perspective, we from Microsoft and Azure in particular, we believe in giving our customers and developers the full choice and flexibility and working in their terms. And towards that, we see huge opportunity around MySQL and we are here to actually provide them the best managed database on Azure platform. And what's a workload situation like at Alibaba Cloud, Jimmy? I think MySQL is the biggest audience provider of the database in Alibaba Cloud. So we support tens of thousands of customers just on MySQL ideas. And it is one of the most stable products in our cloud. So in terms of the new features ever released, it comes new features that people likes. So I think MySQL is still going very strong on the cloud. And even if we develop any variant of MySQL, it still proves MySQL is one of the very stable, one of the best open source databases to start with in the cloud. Thank you. So Vicente, you obviously work on MariaDB server. Is MySQL tired for you? Is MySQL whatsoever? Tired. Tired? Yes. Is it boring, old fashioned? No, so I think... Or is it wired? I think we should recognize that Oracle has done a lot of work towards improving MySQL. However, it always is in the back of my mind that is this development that Oracle is doing always done because they want to develop the server or is it done as a response to the competition that has formed around MySQL? And that is MariaDB. One project that comes to mind is actually my project which I did in 2013. That was SQL standard roles. And MySQL has got this feature, but it's only now recently in MySQL 8.0. So I think it's important that other forks exist because that creates a competitive ecosystem and that people have to keep improving their products. Both Oracle and MariaDB have to keep evolving to keep up with their requirements. So now, I don't think MySQL is tired, but I think it's important to keep everybody on their toes by doing the right thing. Excellent, so it's kind of wired, but thanks to also the branches and forks then. So where is MariaDB used today out of curiosity? If anybody is running Linux, I guess that if you are installing MySQL, most major distributions are actually defaulting to MariaDB. So that's one of the big points that it shows that MariaDB is used not just in simple websites, but also across large systems. There's even the banking sector who's migrated. Also the Wall Street is moving towards MariaDB and pretty much any large scale website ends up using some form of MariaDB behind the scenes. Any of the cloud vendors willing to share a big MySQL or MariaDB win? So I mean, on the manager side, I mean, of course we've seen very, very large wins across the board, like especially running mission critical work, like in the financial sector. HSBC comes to mind that are running thousands of transactions a second and chose MySQL. And yeah, I mean, you're free to go to the azure.com and look at the customer testimonials. So we have a wide, wide range of use cases for MySQL in the financial industry, in the healthcare industry, and of course it is the default database of choice for WordPress, and we all know that. But then we also see opportunities around and we've seen good, big installations of MySQL in the context of Magento. And really when we see where MySQL is used and when we see these opportunities, right, so like Marketo, big user of MySQL, Adobe, big user of MySQL, of course, MySQL is not the only database they use. They use MySQL, they use Postgres, they use multiple databases, and that's the world we live in right now. So it's not about, in this customer, in this label, only MySQL is used, or only MariaDB is used, or only Postgres is used. In fact, we see all these databases being used for different applications and for different purposes. And that's where I think it's actually a goodness. And now that the fact that the developers are really empowered to use whatever favorite databases they want to feel comfortable with and deploy their applications. And that choice and flexibility is what kind of provides the total freedom and use the best database for the right purpose. Right off. Any other? Yeah, we also offer MariaDB in our IDS offering. Another thing to note is Alibaba also invest a lot of money in MariaDB Foundation. So I think we agree that the competition brings the best of the MySQL ecosystem. So we want to keep the competition to be there. Speaking of foundations, on this stage, we have Alibaba who has invested in the MariaDB Foundation and invested in the corporation as well, actually. Microsoft also sponsor of the foundation and even Precona as a sponsor of the foundation. Is there a reason why everyone is sponsoring the foundation? So Alibaba is obviously looking at the foundation as competition, but you've also put money in the company. So is there a good reason there? Basically, I think we want to also help MariaDB to be able to develop its own feature that helps the cloud, the open-source community. I think that's the one big thing that we want to make sure we also contribute financially also in terms of open-source community. That would be good for the whole thing. For Microsoft perspective, we take a very principled approach to how we look at open-source. Of course, there are multiple different lenses that we can use and of course, we come across some of the customers that will say, I prefer MariaDB because I don't want my business to do anything with Oracle and sometimes, it gets into a religious conversation, but there are customers that say that and are genuinely concerned about it. Now, from the foundation perspective itself, it is for Microsoft to put the money where the mouth is and look at sustainability of the open-source projects that we decide to invest and go deep into. And especially around the open-source relational databases, Postgres and MariaDB are the two big bets that we have taken on in the context of how we enable and contribute with the community, partner with the community and ensure that these communities thrive and are able to sustain for the long haul. And that's the spirit behind us investing in the MariaDB Foundation. We've also started contributing code back to MariaDB. It's not as easy to contribute code to Oracle for MySQL and we also see this as a way for us to open channels and innovate in the open, which is our deepest desire and intent and we've already started following through that. This is also the same principle that we applied behind our acquisition in the Postgres space with Citus data. With Citus as an extension, also having an open-source version of it, having made a lot of contributions to the base Postgres as well as Citus as an extension, we believe that was complimentary to the strategy that we are following in terms of making sure that we are deeply committed in the projects that we are highly investing in. Thank you, Sunil. So Peter, why are you re-entrant as a Foundation sponsor? Well, something may not recognize it, but besides having our own software per Kona server for MySQL and per Kona server for MongoDB, we also have a support a lot of customers on MariaDB or on MySQL server. And it felt right for us to make sure that we contribute to MariaDB ecosystem through a sponsoring foundation in addition to the code, what we do. Now, having said that, I would say that I am not super thrilled with MariaDB Foundation in terms of, I would love to see that being more focused on the larger needs of MariaDB community, and kind of being less tight at the hip with MariaDB corporation. And for a long time, this was the reason we did not support MariaDB Foundation because in fact, that would be at large extent funding our competitor. But then in the end, we decided to still go ahead and support MariaDB Foundation. Fair enough. Facebook, are we gonna support the MariaDB Foundation as well or? So, the situation at Facebook, it's pretty different from other companies. So, we are not offering a software service. So, and we are heavily using MySQL 5.6 right now and seeking to migrate to 8.0 in the future. So, we heavily rely on the Oracle based MySQL. So, that's the reason we are not much considering about sponsoring foundation. But we are still having a relationship with MariaDB as specific employees. For example, there is a senior engineer called Sergei Petronia, who was the first big contributor to Myrox, which is a rock-stabby storage engine that is able to use that Facebook. So, we have a relationship with MariaDB for that case-by-case basis. Fair enough. And we'll come back to Myrox in a bit. So, we sent you, you know, who is the steward of MariaDB server and since, you know, carrying on from Peter's discussion. Right. So, I actually wanted to interrupt Peter and bring a comment, but so the- Let's argue. Yes, let's start doing that. That's why they're on the end. So, they can't throw punches. So, the MariaDB foundation is the steward for the MariaDB server. We, it is true that it is in a tricky position in that we share a trademark with the MariaDB corporation. But the thing is that the foundation is the nonprofit and it's goal as I can highlight in our annual report. I have lots of these and I can share it with you. So, our goals are adoption, collaboration, free public access, and to promote the development of the MariaDB server. And I do agree with Peter that we should focus more on community and which is why we are. And I am actually in charge of the team who is going to be reviewing and encouraging more community contributions in the future. We are specifically hiring people to do that. But just to show a quick statistics of what happened before, compared to Oracle, we have four times as many pull requests accepted into the MariaDB server as opposed to how many pull requests Oracle accepted within MySQL. So, we are collaborating with developers, with package maintainers, with distributions as a whole to make sure that the server is working properly, not just the main platforms and distributions, but also the exotic world of, say, ARM, our PC, MIPS. I know ARM is not really that exotic, but it is not something that people immediately think about. So, all these are tasks that Foundation takes care of. In some regards, we do need outside developers to help. Most notably, the corporation provides mostly development resources for the server, and that's fine. We don't want to be the main developer for the server. We want outside entities to come and build the best server that can be. So, the Foundation is the steward and is working on becoming better. Do you think this is a good model going forward, steward-wise? Well, I just want to make a point where my concern is. I think what Vicente just mentioned is what MariaDB Foundation is focused on MariaDB server. But the server is only one component. Same as Linux kernel is important, but not the only one component of your Linux distribution. And the fact what MariaDB Foundation yields all that other stuff to take care of a MariaDB corporation, which often would provide non-open source tools for that, I think that is a really very bad scope for a mission. So, for example, you would see what in the MySQL MariaDB world, where people are standardizing on the proxy SQL to use for their traffic management, right? And it's not just us. You would find, for example, a lot of even Oracle blog posts accepting what that is a very good alternative, right? But I have not seen what MariaDB Foundation, MariaDB Foundation embracing that. And why don't we do that? Well, of course, because there is a BS license mark scale which MariaDB Corporation produces, right? And I believe that as the time goes, you'll have more and more and those tensions created because guess what? MariaDB Corporation is then to find a company which accepted by this time how much? Is it 100 million bucks? Yes, under 100 million. Just under 100 million bucks, right? And those guys, while they may slightly care about open source, they care most and foremost about returns, right? And what that means is what MariaDB Corporation will naturally pursue wherever strategy which we'll have to do, right? To provide the chance of those investments, right? And that is why I do expect what there will be, will be more tensions between the open source MariaDB goals and the commercial goals of MariaDB Corporation. So some of you may have been at Stephen Wally's talk yesterday and he said, Parking your identity brand on any open source project you own instead of the product solution your customers buy creates confusion for your messaging to customers. And the famous example around this is of course how Red Hat managed to have Red Enterprise Linux and Fedora and Vicente has already mentioned that with MariaDB, it's shared and Peter's also talked about the fact that the corporation is heavily venture backed. And so they're likely generally two stewards. Now, Oracle is the steward of MySQL, next slide please. And they actually made promises to acquire MySQL. They made 10 promises, next slide please. And of course they removed it from the internet so it's only on archive.org. And to be fair, they made five year commitments and now it's been more than 10 years. And do you think Oracle, as Oracle killed MySQL or have they stuck by their commitments and are they doing a good job? Yoshi. Yeah, I think they are good. So if they hide the duplicate storage engine in PIS we get screwed but I'm pretty sure Oracle won't do that. I think Oracle is doing pretty good so far. Still lots of engineers like InnoDB or replication teams that continue to work at Oracle so which is a good sign to me at least. Jimmy, you were also part of the InnoDB team. Is Oracle a good steward of MySQL? In general, yes. I worked for Oracle in MySQL. I was in the InnoDB team before they acquired the MySQL team. So it has been like nine years. I just left the Oracle last September. So in general, I think they did a pretty good job. At least leave MySQL alone. Now it's saying that really sponsoring but at least leave MySQL alone and it's developing and it's thriving. So it has one of the biggest in MySQL, InnoDB I think in the world. So they are still getting out a lot of good features that keep the mainstream still very attractive. So everyone wants to move to MySQL 8. That's the latest release. So I think it shows they're still doing a pretty good job. Are you seeing a similar interest as your MySQL? I mean the way I would answer that question is I think if you look at from a business commercial interest standpoint, I think Oracle has done a phenomenal job. And it still garners a lot of revenue for not just Oracle but the entire ecosystem as well. So I think that has really thrived. But if I have to be critical, I would say around the fact that were they able to upheld the open sourceness or preserved the roots of how and why, the reason behind why MySQL became the database it is, then that's perhaps they have not paid much attention to. So I think I've draw the similar kind of parallels like to Peter's point around MariaDB as well between the foundation and the corp. I think there is opportunity there, I believe. There's huge potential and opportunity for MariaDB. And in fact, another foobar database that might come out as well. I mean, the database is never gonna die. I mean, there's gonna be tons and tons of use cases and opportunities for hundreds of databases. But important is staying focused. And I would really like to see MariaDB come out of the shadows of Oracle and MySQL, build their own identity. And also, I think this is for the community at large, once you have a mission, once you have a vision, people align to it. And I see a lack of that alignment, but there is potential, there's huge opportunity. And I truly hope and wish that MariaDB will capture that opportunity and create their own identity which will create a new ecosystem, which will create a new coalition of enthusiast trying to and building support for that. And with our investment behind MariaDB, we believe that we can hope to achieve that. Right off. What do you think, Peter? Have Oracle been good stewards? Well, from technical standpoint, I like Larry's MySQL a lot more than like Monty's MySQL. And I've been out there during MySQL five zero times, right? And this was, I would say, entirely, filled with checksbook features, right? You would have things like views or store procedures or subqueries designed which would not work well together or work well at all, right? To a point, for example, what Google in one of the later releases would have to disable subqueries on the partial level, if you remember that, because they were so bad and they couldn't teach their developers when exactly they are useful, right? Now, I think if you look at the Oracle, they have been doing a lot of solid engineering, right? And some features have been taking its time to come, but they generally come is quite solid, well-documented, well-integrated. Now, speaking about the other side, Oracle, of course, is Oracle and they pursue their open core model for MySQL, so not everything is open source, there is a commercial version, but they have been actually quite nice about that. Wherever they have implemented as a closed source, which is done through plugins, with public APIs, and we at Percona implemented our own open source versions for majority of them. We could do it both from technical standpoint and they did not make it hard and we could also make it from legal standpoint, right? Not getting into any legal trouble with Oracle. So I think what Oracle have been pursuing their own business interests with MySQL Enterprise, but they have allowed the other companies to implement the open source alternatives, which if they choose so, which we did. So speaking of disabled MySQL features, you know, at Facebook you have your own MySQL tree, do you also disable features so that your developers can't shoot themselves in the foot? So we heavily rely on the storage engine API, so we call the new storage engine called MyRogs, and we also rely on the audit plugins, so we have a custom audit plugins that are sending the MySQL query logs to internal queue subs called Scribe, so it uses an audit plugin, so we heavily rely on the plugin development by Oracle and MySQL 8.0 pushes further about plugin or component architectures like MySQL applications, like same thing, so I think this is a very good trend that I hope Oracle continues that. Fair enough, so you know, we sent you, I guess, quality features are important so that people also don't disable features. Is this a focus of MariaDB server, like fully documented, full-on quality features? So we should be fair and say that we still need to work on our documentation, there's things we can improve there, and we are focusing on fixing that. One thing that MariaDB really focuses on is ensuring that your server, which can be from version 5.3 all the way to 10.3, it will always work, you can always upgrade from an old version to a new version, so we care that your application will keep on working even if you were using an older server and then migrated to a newer one. This is something that makes it difficult to do quality features in an efficient manner, so we have to make sure we maintain backwards compatibility all the time. This slows us down, but we believe this is important because users should not be worried of upgrading, and I can make a statement that MariaDB 10.3 is more compatible to MySQL 5.7 than MySQL 8.0 is compatible to MySQL 5.7 when it comes to upgrades. So... Full statement. And I think there are pages that can prove that this is so. Fair enough, so you know, Cloud vendors famously love to disable features of servers, so are you disabling less features on MySQL or are you disabling less features on MariaDB? Thank you. So, it's not about disabling features or not taking the features and putting it in the hands of the customer and the users. As a managed database service, we have to package it in a way that is reliable and not that just works 80% of the time. We have to design it such that it works 100% of the time. And in ability for us to deliver on that promise, we have to take some of the capabilities out of the hands of users because not all users are created equal. And there are some users who can really hurt and bring their databases down and in which case we as a managed database service are kind of reliable for some of those actions. So we take those very seriously and there's a reason why it's not that we think about it as taking features or pulling features away from the users but enable as much as we can provide the guarantees that we are liable for the user that this database is going to be up and running all the time, 24 by seven by 365. So fair enough, not quite saying which one's got more disabled functionality. What about on the Alibaba side? I think similar to what Sonia just said, we are basically a manager of the database for the user and a large part of the cloud service for the database we spend on is on the control system which we provide monitoring and the control and but we cannot release all the configure variables attract to all the users. So something we manage by ourself and that take a lot of, just like Sonia said, they cannot shoot themselves, right? So it's easier for them to use. That's basically the thing. We also provide as many features as we can to the user. So that's mostly the thing that we try to provide as many as we can, but in terms of some dangerous, very tricky configure options, we'll probably will handle ourself or even show it from the user. Okay, next slide please. So, now some spicy stuff. Just last week, Michael Howard, the CEO of MariaDB Corporation said that many cloud vendors, he wouldn't name them but you will know whom apparently, are strip mining open source technologies and companies and abusing the license and privilege and not giving back to the community. And of course, he also believes that proprietary and closed licenses are dead. You're gonna be a general purpose database. Now, what do you think about this, Peter? Well, I will start by saying that it was interesting to see the person talking about open source license dead and then having Clark StrixDB, which as I understand is commercial license and as I hear there is no plan to take it open source, or MariaDB Corporation have a history of taking the open source things which was first with max scale and then taking that to the non open source license. So, on that fact, I think that makes a very good and quotable statement but I don't think that has the truth behind that when it says to in this case. Now, in terms of a strip mining open source and this general concern about the cloud vendor, I think what is very interesting is you hear those concerns from a lot of venture funded companies, right? And I think what is interesting in this case, it looks like the people went ahead and embraced the open source license because of adoption, right? And one of the reasons that adoption with open source is because that doesn't provide their possibility only for one vendor to make the money on the software, right? Allows that to do everybody else. But then as the time comes and as some strong competitor as a cloud vendor arises, they claim, oh, that's not fair. We have to be making sure what we are making all the money from the software. Well, in this case, you guys should have probably not going open source routes, right? Or choose the different license, right? For your software which would have prevented cloud vendors from doing so, right? So, I think in this case, as the open source software works as expected, right? And that is the privilege open source license gives cloud vendors which they use it. Now, another interesting thing in this case is if you don't ask those companies, but if you ask the end users, right? What do you think about that? And then here, you'll find the opposite thing. You will find what a lot of cloud vendors made open source databases much more accessible and ever to use than ever before, right? Especially, I think the rise of Postgres those days is a lot because of the cloud vendors because PostgreSQL often was hard to manage for beginners and then Hiroku and later databases as a service and many major cloud vendors make it not a problem at all, right? Not an issue. So, I think in this case, that there is a lot of benefits even our unnamed cloud vendor provides as a value to the open source ecosystem which is just not in the sense of code at this point. Would they like them to contribute more code? Yes, of course, I would, right? But I don't think what we have, you know some abomination here. So, Jimmy, you represent a cloud vendor. Do you think you're a strip mining open source? I think I don't agree. So, basically, if you look at the cloud vendor when we put a database on the market, a lot of portion of the work is not really database itself. A lot of them are the hardware. A lot of them are the control system, how to make it easier for use to use. So, in terms of cloud, actually it promotes the open source, the MySQL to a lot of customer that don't know how to use them. Now, they make it easier for them to use that. Even for those small customer, most of the 95% of us are small, medium customers. They even, they build themselves, they probably hire somebody and download the open source and do this work. But for us, we just make it easier for them to do that. And a lot of investment is just not in the database itself. It's on the control system, the interface, and also the hardware. So, I think in a sense, the cloud actually make, promotes the open source. It doesn't really, you know, try to strip the revenues from the open source company. Fair enough. What about Yusin, what do you think about this? I mean, like I mentioned earlier, in Microsoft, we've taken a principled approach. And we believe actions speak louder than words. And the acquisition of SIDES on the Postgres side actually echoes our deepest intent and continuing the open source and contributing back to the open source. We are already contributing to MariaDB, of course, in small quantities. But here, strip mining, I don't want to associate with quantity. Even if you contribute one line of code, it's one line of code. It's a contribution. So, we are already walking the talk. And I do worry about strip mining in the harshest sense. And it is, I understand that there is always going to be some commercial interest and some technology that you build as extension that you want to keep for a closed source. But I also believe it is fundamental for the sustainment of the open source projects that vendors contribute. And the contribution can come in multiple flavors. It should not just be measured on a single yardstick of how many lines of code. And really, that's how the community operates. And different companies and different users bring different strengths to the table and as a whole, we create an ecosystem. So, I would also like to see as much as these comments do make airwaves and grab headlines that we really see this as more holistic and not just kind of create our own yardstick and then uphold somebody else on that yardstick and say, hey, they are doing this or they are doing that. And I think this is where partnership, collaborative engagement is very important to the overall growth and the funneling of these things. Sorry, I just want to add one thing that even during the recent, we are importing mysql eight that we like a five, like a six, seven bucks in a month to the upstream. So that's like a one way for us to really test out the open source data so other user can benefit. So that's something that we have the resource to really try out the new things and both in the performance or the code stability side. So that's one thing that we can also contribute making the open source much better. So Vicente, do you prefer code or cash? So before I answer that, gives me a bit more time to think. So I really want to support Sunil and Jimmy and also Facebook Yoshinori. We've gotten significant contributions both from Microsoft, from Alibaba, and we certainly couldn't have gotten my rocks into MariaDB without their help. So this proves that large companies can and do contribute to open source. There are players that are obviously on the other end of the scale. But there are companies that are not just strip mining open source. And now to answer the other question, if I prefer code or cash, why not both? So I suspect largely it's a situation where a foundation likes both code and cash but a corporation probably wants more cash and some will even go up to Facebook and say, look if you've made so many hundred millions this particular quarter, a certain percentage should be going to us since you're running in this very large database. Do you get vendors coming to you and saying that, Yoshinori? So we operate my skin production. So we care much about reliability and efficiency. So for example, if the database fails 10% more often, then our production engineers get burnt out, so which is very problematic for us. And so if we can save the number of servers by 10% that's a huge cost to be in for us. So that's why we care so much about reliability and efficiency, which might be quite different from many small companies. Fair enough. So we've got approximately less than 10 minutes left and I'd like to open up to questions in a while. So in the meantime, we'll have 30 seconds per panelist to tell us something interesting about what you're up to. So Yoshinori, tell us in 30 seconds what Myrox is and why it's important. Okay, so Myrox is a MySQL's new storage engine based on the Rockstay-B. So Rockstay-B is a key value store based on not the B3 but on the rock-structured merge trees, which is similar to Cassandra or HBase. So we created Myrox just for saving space, which means saving the number of servers. So we migrated from InnoRev to Myrox on Facebook.com, the main Facebook database and we migrated from a Facebook messenger from HBase to Myrox, so which eventually saved the number of servers by half compared to InnoRev and we improved reliability quite a lot on the messengers. So it's mostly for saving space and improving reliability. So there's a good chance you all interacted with Myrox today, yesterday, and so forth. What about from a Percona standpoint? Can we get Myrox and Percona? Oh, yes, yeah. Myrox is available in the Percona server and if you want MySQL compatibility plus Myrox, that's really easy way to go. And as Yoshinori said, if you want to stop wasting money, then you should try Myrox. What's unique in Microsoft's offering of MySQL and MariaDB? I mean, so first of all, on the managed database side, we are building on top of the community editions and we do that deliberately, we do that in purpose from our user standpoint that prevents them lock in. And which is what I mentioned before, that we want to serve our customers on their terms and not get them locked into the platform. And once they come, we provide them the most unique value in terms of built-in high availability without having to set up replicas. We are the only cloud that does that and which means you can set up a highly available MySQL or MariaDB database and spend half the cost, even when if you're running on hosting on a virtual machine. That and then providing the enterprise security and enterprise performance. We are working on building and making performance optimization easy and simple. And we call it intelligent performance and it's really about providing users with intelligent insights, recommendations, auto tuning so that they can spend more and more time developing and writing apps as opposed to managing and optimizing databases. Thank you Sunil. Alibaba is famous for also offering MariaDB TX, the enterprise product. Also does offer Max Scale in the cloud, which used to be GPL and now business source license. And you're also developing PolarDB. So I'll give you one minute to tell us why you've gone the enterprise route and tell us more about PolarDB. That's too much favoritism. Okay. So we actually use Max Scale in for our PolarDB as well as RDS. Basically that's used by a surprise by MariaDB. And that use it's doing pretty well. For the PolarDB side, PolarDB is a unique cloud database based on MySQL and it does the physical replication and also separate the computer node from the storage node. So you can scale the computer node or you can scale the cloud node. So that's like a similar to Amazon Arola, but it's different in the design. But it's our new generation database that it's growing very fast for us. And Vicentio, why MariaDB in 30 seconds? Like what's new? So you should choose MariaDB because it's guaranteed to be the server that will always be open source. There is the MariaDB foundation which guarantees that being a non-profit organization. And if you do want to contribute something to MariaDB or you want to make sure that something gets fixed, making sure your voice is heard either via April request or by commenting on our mailing list or knowledge base, you are guaranteed to get an answer from us. You are not just talking to a brick wall. See, I suspect that MySQL as we all talked about is probably still going to remain open source. And Facebook, for example, is a big user of 5.6 and on the path to migration to 8.0. Can you tell us a little bit more about why you're still sticking to MySQL and not MariaDB and what's the path look like for you? So we are using MySQL because personally I want Oracle to continue to support MySQL. So we want to support Oracle as well, but we also have a good relationship with MariaDB and also from technical perspective, so we are like a MySQL GTID and GTID format is different from MySQL and MariaDB, so which makes migration really hard. So migration is hard and you stuck with MySQL GTID. Now why did MariaDB switch to another form of GTID and can people move between both databases? I'm not going to answer that question fully. Just that the GTID implementation from MariaDB was a decision based on sound technical principles. We did not try to be incompatible with MySQL on purpose. If I remember correctly, our GTID implementation was the first one that became stable compared to MySQL, which was a bit later. That's true, it became stable at first in a Percona server release and then might have shifted. That was group commit. So I'm surprised you didn't mention one of the key features which is SQL mode equals Oracle. So are you seeing mass defections of Oracle to MariaDB server now? So it is true that a few large customers from the MariaDB corporation have migrated to Oracle. I don't have access to more users to know if they have, but just seeing a few is probably that a lot more are following and seeing their example. Quite a few people are moving from Oracle to MariaDB. I think the number, the main reason why they switched is that cost wise it's potentially 100 times cheaper to run MariaDB than to run Oracle. Fair enough. Are you seeing such defections, Peter, at Percona? Well, yes, but mainly to Postgres. Yeah, I think what if you look at from the target open source database for migration for Oracle is Postgres is very far ahead from MySQL and MariaDB combined, right? If at least 10x more folks are looking to move to Postgres than to MySQL and MariaDB. So as a last question for all the panelists, if you're going to start on MySQL today and I looked at the manual is 6,028 pages long, what advice would you give to someone starting with MySQL today? What's the quickest advice you can give to someone, Peter? Learn, explain. Learn, explain. Okay, Yoshi. Very high-performance MySQL. Sunil. And SQL support. We also have a lot of wiki blog on MySQL, so we can share that. Okay, we'll send you. Start using it. If you're running into troubles, start asking for help. Okay, so I guess we don't have time for questions. Or do we? Okay, we have time for at least a question. Okay, please. Okay, so the question was, how does Microsoft balances between its own proprietary interest with SQL Server and all the other open source databases? I mean, I believe we have done, like we are doing a phenomenal job there. And that comes with the realization first that Microsoft is building and continues to build a relational database platform. And the platform which supports multiple database engines. And in other words, we are not selling a database allegiance to the user. Our user comes in to us saying, I have these database needs, I want to use MySQL or I want to use SQL or I want to use Postgres or anybody or MariaDB. And we provide them the platform on which they can be successful. So we want to contribute in all, across all, and there's absolutely zero conflict of interest, if you will, in terms of contributing one versus the other. There is no favoritism on contributing more here or contributing more there. Because regardless of the database engines that the user has chosen, we want to be the best platform for them. Okay, and we have one more question. I had a question, which is maybe a follow up on the cloud vendors having to turn off services. Is there more work that needs to be done, like with, I guess, respect to multi-tenancy? Are there other changes that, so when you're dealing, I guess, in a cloud offering, which might not have been the additional kind of design, what might not have been conceived at the time, are there more things that could be made where then you could turn on the features and, I guess, charge for them and not lose the whole system? So it's got less to do with charging more or charging less. It's really the biggest friction of a managed service that we are unable to provide is control on the customization as it relates to the parameter tuning. You could, for example, you could tune the parameter one way that maximizes the performance, but then completely doesn't take into account the durability. Or you could create tables like MySAM, as like a classic example, without, and then lose the table and then lose the data. And it's these type of fundamental key features that we don't give access or control back to the user. Otherwise, from a SQL programming surface area and the language functions, et cetera, they're all fully supported. Like, you have the 100% programming surface area that you would use, like on-premises, on your desktop, laptop, anywhere. Another question? The main reason we changed from MySQL to Postgres was because of the GIS. And I haven't seen you guys mention anything about it, so I'd like to know what's coming ahead for all of you guys. Well, I would say what GIS was very substantially improved in MySQL X, I think it was like at least 10X, a difference between 5.7 and 8. So you can give it another look if you are interested in this space. And MariaDB has said GIS for quite some time as well, possibly even before MySQL did. Yes, I'm not really that familiar with the GIS system. Yeah, so it's been around since 5.3 and got improved further. As MySQL implemented GIS functionality, MariaDB also then follows on. So this push actually goes both ways, right? So to make it more compatible, like GeoJSON and stuff has come into MariaDB as well. So we do have a dedicated developer working just on GIS and improving it, so we do care about it. I guess one last question from the internet. It seems like data stores as of late have an increasing trend of changing their licenses to things that are less than... To other OSI licenses. Is there something that makes databases specifically a target for this? I mean, there's folks on stage that have done this as well with some of their product. And is there some other way that we should be looking to make this a more sustainable space other than playing with the licensing? So my personal opinion on licensing is that it's usually VC-backed companies that play with it. And after MySQL got sold for a billion dollars, it turns out lots of VCs-funded databases, right? They're probably like 100-odd databases that have been funded. Not all are gonna survive. I mean, many have passed away recently as well. And it's a way to drive revenue. The good news is just this weekend at scale, MongoDB actually withdrew the SSPL license from trying to be approved at the OSI, which is interesting. So they're actually not going to be a proprietary database, which maybe makes it fairly interesting for even you, Peter, because you actually make MongoDB product. Well, yes, yes. I would say that is a very unfortunate choice for MongoDB, right? If you would be at this choice, now we would not have added to MongoDB portfolio. But now we always have a lot of customers and they made commitments to them. And that means we'll continue to develop your corner server for MongoDB and SSPL license, even though it is not open source, but we don't have any choice. At the same time, what we are looking for is the good open source and not open source alternatives for MongoDB users, because again, competition makes things better. I'm very glad to see, for example, there is Amazon now has a document DB as a comparability, right? I am very glad Microsoft has a Cosmos DB, MongoDB comparability, even though those solutions are not open source. I am even more excited we have a foundation DB, document layer, right, which is 100% open source. It's not 100% compatible, yes, with MongoDB. But I believe with those variety, right, we will see more alternatives, truly open source of alternatives for MongoDB merge. And Alibaba, of course, famously also has a MongoDB service to be fair. Yeah, I think we were kind of worried about the, well, pay very attention to this license issue. And that also, I think one of the reasons a lot of people using Postgres also is because the license issue. So I guess this, unfortunately, is some kind of issue that we all pay very much attention to. And I suspect the good news here is that MySQL and MariaDB server are GPL v2 licensed. And it's fairly strong with lots of external contributors. Getting them to re-license code would be a lot harder. So I think MySQL is going to be GPL v2 for the long haul, as is MariaDB server. All right, thank you very much for listening. Thank you, all panelists. Thank you. Thank you.