PlanetScale, the serverless MySQL database built atop Vitess, the open source project developed at YouTube, aims to help developers be more productive with CNCF technologies without needing to be experts of all of them. The company announced three new features at KubeCon: PlanetScale Insights, PlanetScale Portals, and PlanetScale Rewind. PlanetScale Insights is the beginning of an APM stack inside of PlanetScale, enabling users detailed database performance metrics and to identify queries within seconds. The PlanetScale Portals feature has the ability to project read-only replicas of your database into any AWS region that PlanetScale is in today. The company’s third new feature, PlanetScale Rewind, is an ‘Easy Button’ to undo schema migrations that enables recovery in seconds from bad schema changes.
“You can have the best schema change process, the best schema change tool in the world, but if you, the user, are typing in a bad schema, it doesn’t matter. All those tools get you to the cliff faster,” says Nick Van Wiggeren, VP of Engineering at PlanetScale, on this episode of TFiR Let’s Talk, at KubeCon in Valencia, Spain.
Key highlights in this video interview are:
- Van Wiggeren discusses the history of PlanetScale since its conception four years ago and how it is making Vitess and CNCF technologies more accessible to developers.
- PlanetScale is helping people pick the database that will get them to hyperscale as early as possible, easing them all along that journey. Van Wiggeren explains the challenges companies are facing with scalability and the use cases they are seeing.
- Van Wiggeren discusses the role open source technologies play in PlanetScale and challenges of building out features specific to certain businesses. He discusses how the company balances open source collaboration with the needs of commercialization.
- PlanetScale has announced the release of three new features: PlanetScale Insights, PlanetScale Portals, and PlanetScale Rewind. Van Wiggeren explains the key details of the new features and what challenges they are solving.
- PlanetScale is launching two new pieces of content: a video and blog post on the future database, and a podcast on the history of the database. Van Wiggeren elaborates on what will be covered in the content.
The summary of the show is written by Emily Nicholls.
Here is the automated and unedited transcript of the recording. Please note that the transcript has not been edited or reviewed.
Swapnil Bhartiya: Hi, this is host Swapnil Bhartiya and welcome to another episode of TFiR: Let’s Talk, at KubeCon in Valencia, Spain. And today we have with us, Nick Van Wiggeren, you are VP of engineering at PlanetScale. First of all, welcome to the show.
Nicholas Van Wiggeren: Thank you, pleasure to be here.
Swapnil Bhartiya: Yeah. And of course, for me, it’s the first day, but it’s couple of days of this event.
Nicholas Van Wiggeren: Yeah.
Swapnil Bhartiya: I’m pretty sure that you folks have been to the keynotes, you have been to the booth section, also. Tell us, first of all, what kind of energy you are seeing because this is the first KubeCon of the year and also, in person for a very long time.
Nicholas Van Wiggeren: I can’t tell you how fantastic it is to be back in person, seeing booths, seeing people, gathering new information and even just the kind of the breakout sessions and stuff like that were happening yesterday. I didn’t participate in any, but I peeked my head in and there’s just hundreds of people heads down on laptops, collaborating together in person. It really feels like things are really getting back into the swing of things, it’s great.
Swapnil Bhartiya: Yeah, and just an anecdote that KubeCon is also where I met Nick and I learned about PlanetScale. So, that’s also kind of little history there, so since the name of the company is already here. So, as I said, KubeCon is where I met PlanetScale. So talk about, first of all, the whole journey of the company itself in couple of years.
Nicholas Van Wiggeren: Yeah, that’s a great question. So, just a little bit of background on PlanetScale. We are built on top of Vitess and we’ve been around for a handful of years, three, four years now. And what we’ve done is, we’ve taken Vitess and built a serverless database on top of it. So, around the middle of the year before last, we kind of started scratch with a brand new kind of control panel, web interface and things like that.
And middle of last year, about summer of last year, we launched the beta of that product, so it’s a serverless database, MySQL-compatible with Vitess under the hood, but what we added was a whole series of schema workflows so that you can easily do schema changes, collaborate with your team, with your coworkers on those schema changes, think of it like pull requests, but for schema changes, and that went generally available in November. And since then, we’ve been adding features and kind of building out the whole database platform. So, we’re built on top of Vitess, we love Vitess and all these CNCF technologies, but what we want to do also is, add context to them and make it so that people can be productive with them in companies and don’t necessarily have to become experts of all the underlying technologies.
Swapnil Bhartiya: If you look at Vitess, it was trying to solve a big problem for YouTube-
Nicholas Van Wiggeren: Exactly.
Swapnil Bhartiya: But if you look at today’s world, when we look at cloud, their companies, of course, hyper killers, major companies who are using Kubernetes. And then there are small companies who are using Kubernetes, in more or less like Linux, it has become Linux on the cloud work.
Nicholas Van Wiggeren: I completely agree with you.
Swapnil Bhartiya: Which also means that the problems that are folks facing is different. So then, we just like Zoom, because we are sitting within a camera, just, Zoom into video itself, what kind of problem you are trying to solve when it comes to scalability and what kind of use cases you handle.
Nicholas Van Wiggeren: Great question. So, I think you really raise a couple of good points. I like to talk about layers of abstraction, right. At first it was hardware, we added Linux and operating systems on top. And now you don’t really think about the operating system that much, you know how to operate inside of it. We added containers, which created a light abstraction after LXC, we’ve layered Kubernetes, so now your abstraction is multi machine. And now we’re really talking about another layer of abstraction on top of that. So Vitess was created to solve kind of outrageous scale of super huge websites, right. Think YouTube, things like that, where you’ve got tens of thousands of servers, millions of users, millions of connections. And it’s very good at that, right? It’s been tested at that scale, part of what we want to do at PlanetScale is actually bring that massive scale down to a small size that is cost effective, high performance, but also helps people get started with that hyperscale technology as early as possible.
The database world is kind of fraught, I think, with so many of these, “Oh, you know you’re going to hit a scaling point here.” People are planning for their next database when they adopt their first one, if they’ve kind of been through the ringer a couple of times, so we want to help people pick the database that’s going to get them to hyperscale as early as possible and ease them all along that journey from single database cluster, all the way to 64, 256, or thousand 24 charts of data.
Swapnil Bhartiya: Excellent. And once again, one more point that you made was, to as PlanetScale health companies, that one more thing that happens with the open source is that open source, and I may be totally wrong about that, kind of solve day one problem. But day two is the real challenge starts, where not only you have manage update, maintain, but also add features functionality. Sometimes the community may not need feature, but a customer will. So that’s what commercialization of open source play a big role, so can you also talk about that aspect of credit scale as well..
Nicholas Van Wiggeren: I’d love to, so I think that they’re… I mean, and again, open source technology, we would not have our business without all of the amazing work that’s gone into everything from Linux to Kubernetes, Thanos, Prometheus, you name it, right. That we’re built on a whole host of these open source technologies, and what they do really well is, like you said, solve the day one problem. But what happens at day two is that you need projects that have opinions and you need projects that have features that are specific to businesses. And if open source projects followed everyone’s opinion and matched everyone’s features, they would have 10,000 configuration parameters. And they’d be stuck in this sludge of time for another one time for another one. So what we’re able to do at PlanetScale is start to solve those day two problems by having an opinion, this is how the database should work, this is how the database should change, this is how Kubernetes, a multi-cloud or a multi-region Kubernetes deployment should look like.
And by taking on that complexity and opinion ourselves, we can actually solve business problems for our customers instead of just solving technical problems for our customers and help them build faster and build better for their entire journey. So, nothing we do would be possible without open source, but at some point you have to kind of take an opinion, dig in and say, “This is what we support, if you can match our opinions, you can build a business on what we’re creating.”
Swapnil Bhartiya: A couple of announcements also are coming out this week. And, it’s our partnership, it’s about new features, it’s also about past city that you folks are going to announce. So, one at a time, let’s talk about the new feature that you’re going to add.
Nicholas Van Wiggeren: Yeah. So we’ve got three new features that we’re launching this week and I’ll take you through them all very quickly and we can kind of go through them. So the first one, the one I’m most excited about is called PlanetScale Insights. And what Insights is, is the beginnings of an APM or an application monitoring stack inside of PlanetScale. So oftentimes we see people… They use a database, whether it’s an open source database they host themselves, a cloud database or anything like that. And they start to use it and they look and say, “Okay, well, why is my database not performing? Is it my queries? Is it my database? Do I need more resources?” And then you have to go buy or build or install a whole nother piece of software to get that kind of level of insight, right?
Databases aren’t telling you this by default. So, because we use Vitess and because we have this rich feed of every query, we’re able to say, “Okay, we can actually track individual query performance.” Using things like query tags, you can actually instrument your application, have your application feed in the context into PlanetScale. And we can give you a dashboard that says, “Hey, this query tag or this API route had the top five slowest queries in this hour.” If that API route is too slow, here’s why. So it’s the beginning of what we want to say is the kind of zero configuration, APM stack built right into PlanetScale from day one, so that you don’t have to go, “Oh, crap, I need this.” And then install it and then hope a problem happens again, but you have that history right there built into the tool.
Swapnil Bhartiya: Excellent, next one.
Nicholas Van Wiggeren: Yeah. So the next one is our initial multi-region journey. So we’re calling it, PlanetScale Portals. And what it is, is the ability to project read only replicas of your database into any AWS region that PlanetScale is in today. So if you’ve got users in Australia or users in Europe, users in India, you want to be able to serve traffic faster to them. You can put your application there, using kind of any one of the front end platforms that are kind of globe spanning, and you can also put a read only database right there next to them for super fast kind of local API requests or local web requests.
Swapnil Bhartiya: And then, where do we go?
Nicholas Van Wiggeren: And the third is what we’re calling, PlanetScale Connect. PlanetScale Connect, I’m really excited about, it’s the beginning of building an ETL or extract transform load system into PlanetScale. So we are working with Airbyte, the kind of open source ETL tool. And we are letting users do kind of nearly live syncing from their PlanetScale database into any one of the data lakes, data warehouses, like Redshift, BigQuery, or the open source equivalence of those. So their data teams or their marketing teams or their product teams can get nearly real time insight into what’s going on in their databases and can start to contextualize some of that data. So, we see people right now doing full database dumps every night and these kinds of things like that. And we want to help get those business teams the really up to date insights. So, PlanetScale Connect will allow people to plug right into the database and again, turn it into a platform from just being a database.
Swapnil Bhartiya: Excellent, thanks for explaining this feature. Now, another thing is partnership.
Nicholas Van Wiggeren: Yeah.
Swapnil Bhartiya: Talk about that.
Nicholas Van Wiggeren: Yeah, so we’re also launching a DataDog integration. So, we just talked about PlanetScale Insights and a built in APM, but we understand that it’s important for us to be where we are, where our customers are as well. So, you’ll be able to plug in a Datadog API key in and immediately get statistics about your databases size, query performance and things like that sent right over to your Datadog account. So that you can add alerts and help capacity plan, and tie it in with your existing ecosystem, if you’re a Datadog customer already.
Swapnil Bhartiya: Excellent. Now the last piece of-
Nicholas Van Wiggeren: Yeah, that’s all, it’s a big week.
Swapnil Bhartiya: Yeah.
Nicholas Van Wiggeren: It’s a big week.
Swapnil Bhartiya: It’s a podcast series, tell us about that.
Nicholas Van Wiggeren: So we actually have two pieces of content we’re launching. One of them is the history of the database podcast, and when we sat down and talked about this, I was originally a little skeptical, I won’t lie. But after seeing the first couple of episodes and listening to what we were going to make, it’s fantastic. So, we’re bringing in experts from kind of across the field, and we’re going to tell the stories of how the modern database got to be what it is. So starting with literal stone tablets, that people were etching, going into the first kind of modern databases for things like flight search and travel. We’re going to kind of walk people through why the relational database looks the way that it does, and maybe why they have a love-hate relationship with their database today. It’s really good content, and I think it’s going to be a really interesting way of exploring, people don’t think about this kind of thing, right? Why are computers the way they are? We’re going to help add some of that context for the database.
Swapnil Bhartiya: Excellent, is that all?
Nicholas Van Wiggeren: One more. So we also have what we’re calling, the future database. And this is a video and a blog post, and we’ll continue to kind of keep touching on this as we go, but it’s kind of a bit of a calling our shot as it we’re like in pool, where we’re talking about what the database of the future is going to look like. And this is not a one-to-one mapping of PlanetScale’s features, this is not a spec sheet, there are things on this list that we don’t do today, but the idea is to start to call out if you’re technology decider, or if you’re looking to the future and planning, what are you wanting out of your database?
Things like linear scalability, things like being able to query it from anywhere, having it be global accessible on everything from toasters, all the way over to, big gigantic bare metal servers and the kind of thing that are going to get people to be able to develop the next wave of applications. We want to help people think about what they should be looking for in their databases they’re picking, and we’ve looked at our competition, we’ve looked at existing databases and again, we’re not trying to tie this to PlanetScale direct roadmap. We’re really saying, we’re on the cusp, I think, of a next evolution of what databases are and what people are doing with their databases. And we want to help kind of spur some thoughts to help people think about, “Okay, what is my next database going to look like?”
Swapnil Bhartiya: Excellent, is that all?
Nicholas Van Wiggeren: That’s all, yeah.
Swapnil Bhartiya: Perfect. Beyond this announcement, is there anything else that you want to touch upon or do you think that we have talked about the company, we talked about the history of the company a bit, we talked about the larger problem that you focus solving. We talked about the business side of it, and we also talked about this announcement. Is there anything else?
Nicholas Van Wiggeren: I’d like to talk about one more thing, if you don’t mind. So, in March, a couple of months ago, we launched a feature called, PlanetScale Rewind. And what that is, is a schema migration undue button. So you can move your schema forward, right? You can add a table, drop a column, change an index or something like that. What we realized was a significant portion of the downtime that we have taken at other jobs, kind of in our careers, was because of bad schema changes. And you can have the best schema change process, the best schema change tool in the world, but if you, the user, are typing in bad schema, it doesn’t matter. All those tools get you is to the cliff faster, so what we wanted to do was we wanted to give people an easy way out instead of having to restore from a backup or get a time delayed database replica and fast forward it to the moment, and then add the old data back on top.
We just give you a rewind button. So if you do a bad scheme of change, if you remove a column or add a table that you don’t like, for 30 minutes after the deploy request, you can just click rewind and it will swap you back to the old schema, without losing the data that you’ve written in the intermediary portion.
So if you’ve had new user signups, you don’t have to worry about having to ask those users to re-register or going and replay all of that traffic, it will all just work. And this is, again, part of that, I think outcome driven, business driven attitude that we’ve taken toward databases. We’re coming at it from the perspective of a user, from the perspective of a company and saying, “What slows us down? What stops us from hitting our business objectives? Features like this.” So, that’s all I want to talk about. I really appreciate it, we’ve met a lot this week, that’s that’s happening and I’m really excited to share it all. But I wanted to make sure we covered everything because there’s a lot of hard work that’s gone on behind the scenes to make it possible.
Swapnil Bhartiya: I’m not surprised that your name is all of PlanetScale. So, whatever you do is PlanetScale.
Nicholas Van Wiggeren: Exactly, we try.
Swapnil Bhartiya: Nick, thank you so much-
Nicholas Van Wiggeren: Thank you.
Swapnil Bhartiya: For taking time and sit down with me today. And of course not only share this announcement, but also share some insights that-
Nicholas Van Wiggeren: Anytime.
Swapnil Bhartiya: I appreciate your time. And as you know, we should do it more often, depending on whether it’s online or in person.
Nicholas Van Wiggeren: Anytime, thank you so much. Have a great rest of your day, enjoy KubeCon.
Swapnil Bhartiya: Yeah, perfect. We got a cut, let’s stop the recording.
Nicholas Van Wiggeren: Beautiful.
Swapnil Bhartiya: Perfect, okay.