Data Council 2025: The Databases Track with Sai Krishna Srirampur and Craig Kerstiens

• 

  Craig KerstiensProduct and Eng
  Crunchy Data
• 

  Sai Krishna SrirampurProduct Leader
  ClickHouse

Heavybit is thrilled to be sponsoring Data Council 2025, and we invite you to join us in Oakland from Apr 22-24 to experience 3 days of cutting-edge technical talks from the brightest minds in AI & data, learn more at datacouncil.ai and use the code HEAVYBIT20 for 20% off.

Introduction and Sai’s Journey into Databases

Craig Kerstiens: So I'm Craig Kerstiens, Chief Product Officer at Crunchy Data. I've got with me Sai who I'm actually really excited to talk to because I've known Sai for quite a while, since he was first in the industry.

Sai's seen and done a lot, and now is part of ClickHouse, helping people scale data warehouse loads.

But I'm really excited with the track that he's created, but I wanted to kind of, you know...

Sai, don't let me take away your introduction and, you know, give your own background if I missed a bunch of things.

But once you kind of give some of that, I'm really curious to drill into, first, something I've ever asked you, what drove you to databases in the first place?

Like, how did you end up in the data world? What brought you into it? So first, if you want to add anything to your own background and then, like, why database is for you personally.

Sai Krishna Srirampur: Thanks, Craig. So just to start with my background, right?

So yeah, hi folks, I'm very excited to be here, and thanks Heavybit for hosting this great podcast.

And I'm Sai and currently, I work at ClickHouse and I lead database integration efforts at ClickHouse.

So what my team does is we are responsible for providing a magical experience for customers to integrate external data stores with ClickHouse.

Prior to ClickHouse, I was the CEO and co-founder at PeerDB. And at PeerDB, we are building a data replication tool with laser focus on Postgres.

So our unique insight was we were really focused on Postgres and we were making it very easy for customers to move data from Postgres to external data stores.

And one of the data store was ClickHouse, and we added that connector and that took off. And that is why the acquisition happened.

And prior to that, Citus and Microsoft were really transformative in my life because that's where I got introduced to Postgres and that's what instilled passion in me of databases, right? Like, I got to work with a bunch of customers, right?

And in Citus, it was mostly SMBs, like, startups, right? And myself and Craig, like I would say that we sold Citus Cloud together, right?

And at Microsoft, it was a counterpart, we were working with Fortune 400 companies who were running Postgres at scale, right? They were running Postgres at hundreds of terabytes.

I feel that like Citus and Microsoft were transformative in my career because that's where I could see how customers are using Postgres and what are the problems that they were running into, right? And that's where PeerDB came from.

So what I observed was that moving data in and out of Postgres was a big problem, right?

Let it be related to migrations, right? Or let it be moving data to data warehouses, right? So that experience was not very performant.

It was not very scalable because existing data movement and ETL tools were all generalized, right?

So they were not laser-focused on Postgres, right?

So that was the unique insight that we identified and that's the reason we started PeerDB with this vision of building a laser-focused, like, data movement tool for Postgres.

But that's my journey in databases and, like, very excited to be here and be hosting the databases track for Data Council.

Craig: I love that the, like, the Unix philosophy of small sharp tools, right? Instead of being generalized, right, really focused on replication and solving it perfectly from one system to another.

I love that. But, like, how did you even get into databases in the first place?

Sai: Yeah, that's a great question. So the way I got into Citus-- that was my first introduction to databases and Postgres, it was just by chance.

I did not think of getting into databases. I t was my second job out of school.

It was just that one of my great, like, friends who was working there, so I thought it's okay and apply.

Craig: So it was a completely, like, a happy accident or unhappy accident, which...

Sai: Absolutely happy accident, right? And I'm not leaving Postgres and databases since then.

So that's how I got into databases. And my passion for databases continues because I believe that, like, number one, databases are very critical and important to this world, right?

Every application on the internet requires a database. So that is number one. And number two is, as application needs evolve, right, like, database also needs to evolve, and the problems and challenges that it surfaces are going to get only harder. And that's where you need solutions, right? Like, and that's what, like, drives me to be so passionate about databases.

Craig: Yeah, so I'd love... Like, I always say, like, working with databases, it took me personally a long time to get used to this that it's a very thankless job.

It's kind of, like, you don't show up to your bank and say thank you for not, like, losing my money or, like, thank you for, like, not having, like, my social security number stolen, right?

Like, when you hear from your bank, it's like, "Uh-oh, here's a year free of, like, free credit protection," right? Because there was a security breach.

Like, it's really a thankless jobs 'cause you store the data, you do it well, and your customers don't talk to you, right?

Like, it's critical, but it's a very thankless behind the scenes. Like, if you're doing your job well as a database, right, you kind of fade away into the background in what application developers build.

So it took me, personally, a long time to get to that. So in a minute, I want to get more into the track itself, but I'm kind of curious from your journey.

Before we get into that, any more, like, personal insights and experiences and kind of, you know, aha moments that you've had from your time in the data world.

Philosophy and Key Insights on Databases

Sai: I think I have, like, two insights that I wanted to share, right? Like, and maybe you'll agree with one of them, but, like, let's see.

So the first insight that I wanted to share is that across my decade of experience in databases, I believe that purpose built databases are more impactful than generalized databases.

And this comes from my experience at Citus, right? Like, what we observed was we were, like, trying to do, like, multiple use cases, right, like, it'd be multi-tenant SaaS, like, which was more OLTP and transactional heavy, then you had, like, you know, real time analytics, right?

So what I felt was because of our focus on multiple use cases, we spread thin, right? Like, so sure Citus was a great success, right, but it did not become a Snowflake, right?

Craig: I'm curious if this is the one I agree with or the other one because I mean, being a Postgres guy, right, like, Postgres can do a lot, but sorry, keep going.

Sai: Yeah, I think this is the one I'm not sure whether you'd agree with, but I am a strong believer of purpose built databases.

And that's the reason probably I'm working for ClickHouse now and the acquisition went through because ClickHouse is laser-focused on, like, analytics, right?

So that is one of my, like, insights. The second insight, I'm sure you'll agree, is users buy databases not for just the technology and features, they buy databases for the team, the thought leadership, and the support, right?

Like, I think, that is very, very important. And I would say that is equally important as the product and the features, right? So those are my two insights.

Craig: Yeah, no, I think, like, you, imply this, that the roadmap, the direction, the vision, the ethos, like, what a database is and becoming, right?

So when you say to the team, right, where is it driving, where is it headed to, right? What's the vision, what's the direction? So yeah, I'm highly aligned to that.

The first one we might have to dispute a little bit as I think, like, personally, right, I am in the Postgres camp. I've been in that world for a long time and think it's an amazing database.

And you know, I remember, the way, way back the Postgres gray-beards were, like, they remember when XML databases were going to take over the world.

Like, legit this was a thing of XML databases were going to take over the world and Postgres just added an XML, you know, data type.

It went and added JSON, et cetera. So, you know, I'll get off my Postgres bandwagon. I do think there is something to very specialized, you know?

I like the Unix tool philosophy, but I also see, you know, both expansion and compression within the data world.

Vision Behind the Databases Track

Shifting a little onto the track itself, 'cause when I saw the lineup, I was super excited.

It's like, you know, there's one track, but there's a lot going on in here. What was kind of your vision and goal with creating this track?

Sai: Yeah, so the team and vision of the databases track revolves around three main aspects.

First is we wanted to capture most of the current and latest innovations that are happening in the space. That is number one. Number two is we wanted to bring a holistic view and a diverse perspective to the databases track, right? And that is reflected via other speakers and the talks.

Like, and the database track is going to get started with, like, Andy Pavlo's talk, right? Like, and I feel that Andy Pavlo made databases cool, basically, right?

Craig: Yeah, I mean, can we say, like, databases aren't supposed to be sexy. Like, is it fair... Like, sorry, they're not a sexy technology and yet Andy's done an amazing job of that.

So yeah, I mean, right there, you lead it off with kind of the pinnacle of now databases are cool again.

Sai: Absolutely, absolutely. And then we have, like, a bunch of talks, like, starting with, like, transactional workloads, right?

Like, you have talks about Postgres and MongoDB. In the analytical side, you have talks about, like, InfluxDB, ClickHouse, DuckDB, and how the open data format and data lakes are affecting the data warehouse landscape.

There is a very interesting talk by Marco from Crunchy, you would know him a lot, on converging database architectures, right?

Where he talks about how they're extending Postgres for analytics by plugging in, like, DuckDB, right?

There is also, like, a talk by Etienne who is, like, the co-founder of Weaviate, which is a very famous vector database.

And then this talk is about agentic databases, right, and how they are going to be the next big thing, right? Like, this is more related to AI, right?

So that is the second thing where we wanted to bring diverse and holistic perspectives to the track.

Last, but not the least, this is something that is very personal to me, is we wanted to make it really customer-focused, right? Like, so that is reflected by the speakers, right?

Like, most of the speakers are customer-focused, like, practitioners. If you see, like, Paul Dix from InfluxDB and InfluxDB has very large adoption in the community.

If you see, like, Tanya who is the head of product at ClickHouse and ClickHouse has, like, you know, thousands of, like, businesses using it, right?

You have, like, Frank from, like, MongoDB and Frank has, like, a implementations background, right?

So those are the three main, like, things. One is, like, capturing, like, current innovations in the database space.

Next is bringing, like, holistic and diverse perspective to it. And the third is keeping it very grounded and, like, customer-focused, Craig.

Craig: Yeah, no, I love how it's kind of weaved the mix of...

Often you find academic conferences, you find, you know, industry conferences, you find, you know, vendor conferences and it's kind of a gorgeous mix between where you kind of run the gambit all in a single track.

What Makes This Track (and Data Council) Unique

Backing up, like, data conferences are out there. There's a lot of conferences, like, how is this unique from others?

Because I look at it and it jumps out to me. But I want to see, like, from your perspective as the track organizer, how is this unique from other data conferences out there that you've been to or attended or, you know, seen?

How is it different, 'cause to me it feels different looking at it. Like, as part of creating it, can you talk a little bit about that?

Sai: Yeah, that's a great question, right?

So the one thing what I've observed about, like, database conferences is that most of the database conferences are, like, tied to a technology, right?

You have, like, Postgres conferences, you have, like, MongoDB conferences, right?

So most of the times they are tied to a technology, but then the beauty of Data Council is it lets multiple database technologies talk about what is going on in their space, right?

So that is the theme also of the databases track where we, as I said, we wanted to keep it very diverse and bring in holistic views across these technologies. So that is one big difference, I would say, where it is not tied to a single technology, but it's more tied to what is going on in this space.

So that's the big difference.

Craig: The entire conference looks super exciting to me. But I also have to say, I might just camp out in your tracks room the entire time.

Like, are you sort of competing with, like, a... " No, I want this room to be packed versus all the others."

Like, are you camped in this room the entire time and you think it's the best room?

Can you talk a little bit of, like, you know, the other spread of the talks and tracks at the conference?

Sai: Absolutely, there are a bunch of, like, tracks, right? Like, there is the data engineering track, which is what, like, Data Council is known for, right?

I mean that's where it started from, right? Like, and this is the first time we added the databases track because this was a conversation with me and Pete last time.

Like, I told him that, like, "We should talk more about Postgres." I mean that's what I told because I'm a big Postgres fan.

Craig: All right, you've won me back over. Keep going, keep going.

Sai: I am a Postgres fan. So I told him that we should talk more about Postgres because people want to learn about Postgres, which is where Pete, like, had this brilliant idea that we'll start the databases track, but then that you have the data engineering track, there is foundation models track, right?

Like, there's an AI track, right? Like, so there are a bunch of tracks, and I think all of them are going to be, like, amazing, right?

And if you look at the speakers, right? Like, in one of the tracks, we also have, like, you know, the creator of, like, DuckDB who's going to be speaking.

Craig: Hannes, yeah.

Sai: Exactly, so I think all of the tracks are going to be fun, but I would say that, like, you know, the databases track probably would be the best because we have a legendary lineup, right?

Craig: I'm really excited, I may stop out to catch Hannes' talk. I love Hannes' talks and what they're doing with DuckDB.

So there's a lot of, you know, data conferences out there. We've covered how your're different, and in particular, this track among databases, right, from academic to practitioner to, you know, interesting things happening and where the future is.

Audience, Flow, and Accessibility

Can you talk a little bit about the themes and kind of how this flows from one to the next? Because it seems like, it really does flow from a...

You can start your day in this track and end your day in it, and there's a lovely journey five-hour movie that happens, right?

Sai: Absolutely. I think the way I look at it, right, like, so far we spoke about, like, you know, the speakers and the talks, right?

Like, getting into the audience, right, like, who would benefit from this track, right? Like, I did give, like, a thought for this.

And I would say that I see three types of audiences benefit from this track.

Number one is basically database users who want to figure out, like, their data stack and you know, they learn about all the innovation that's going on from legendary experts, right?

Like, and then through that, like, they fuel how their data stack is going to look like. So that is number one. The second type of audience I see is, like, database developers, right?

Like, if you see two to three of the talks in the track are related to, like, database development, right? Like, these are, like, builders, like, who are giving talks.

So these database developers can learn how, you know, actual development is done where these are builders who are building large-scale databases, right, databases that are supporting thousands of customers.

So that is the second audience. The third is, it's also for students, right, like, who want to learn about databases, right?

So I see these three type of audiences, database users, database developers, and students, who would benefit from this track. What I expect is because of the diverse set of audience and the diverse set of speakers, there would be a lot of learning and great discussions. And that's the goal of the track.

Craig: So if you don't want to accidentally stumble into databases, but be intentional about it, you jump right into this.

Like, I agree as a student, right, like, you can go to a lot of academic, just the spread of academic to protect practitioner here is truly unique.

Now, I guess shifting a little bit, was there anything, like, intimidating about creating this track, right?

Like, to me there's a lovely cohesion, but is there any part where, like, "Hey, I'm a Mongo person, this part's going to scare me."

Or, like, should anyone be intimidated by, you know, "Well, this talk isn't for me, that's talk's not for me."

Like, if it's foreign to me, how should I hang out or think about it? Or do you have thoughts on that cohesion versus, you know, how you struggled with creating the track?

Sai: I think that's a great question. So firstly, like, I think it was a struggle to create that track.

The reason is, like, curating all the hundreds of talks and choosing just six of them was very difficult, right?

Like, and we did it, like, more top to bottom, right? Like, where we decided, like, what's the vision of the track?

And then, like, you know, we created these buckets of different topics and then we got the best speakers in each of them.

Second is, like, I don't think that, like, it's going to be a challenge for someone who is foreign to something, right?

Like, for example, agentic databases are not my territory, right?

Like, it's completely new for myself, but then I'm so excited about it because I'm expecting, like, all the speakers in the track would give context on, like, you know, each of their, like, territory, right?

Like, I mean their topics. So I don't see it to be, like, you know, hard for, like, folks who are foreign to a topic because all of these are, like, super experienced speakers, right?

And I'm sure that they would do a great job. So, yeah.

Craig: So we've talked a lot about the track and the flow and the technology.

My favorite part of conferences is actually, usually the hallway track, right? And while I'm really excited to kind of camp myself in the track room and go from, like, Andy to time series to analytical, to transactional, like, just that whole flow, right?

My favorite part is usually the hallway track and connecting with people. Like, can you tell me a little bit of, like, the community you want to kind of want to bring together as part of that?

Sai: Yeah, so that's a great question.

I would say that one important goal of the track from the hallway and from the audience perspective is to keep it full of questions and keep it very conversational in the sense that the audience can ask questions, let it be during the talk or let it be off their talk, right?

Like, because finally, like, even though there is a legendary lineup, right, like, all of these speakers are just, like, people, right?

Like, I'm sure that they will also be learning a lot, like, from, like, the questions, right? Like, so I think it's going to be very important to keep it very conversational and, like, keep it very questions heavy.

And also not to forget, the audience can be the next speaker, right? So I would really recommend folks to submit talks.

Like, this time, they were, like, hundreds of talks, right? Like, and it was so hard for, like, the...

Craig: So you want your job to be harder next. You want more talks submitted next year and a harder job is what you're saying.

Sai: I agree, yes.

Craig: No, I love it. No, I think it's spot on. Like, you know, having spoken at conferences, right, like, there's this aura about you, but, like, no, like, many people in the audience...

I actually once gave a talk and in the back of the audience, and it was the most intimidating talk I ever gave, was Michael Stonebraker, the creator of Postgres.

And I was giving a talk on Postgres, and there were questions that came up and I was just, like, sweating bullets. 'cause here's the man that created Postgres that's won a Turing award, right?

And I'm giving a talk about Postgres, about the thing that he created and I had no idea. I just, the last minute, saw him walk in the back.

And so your audience can be full of experts. And so I think, you know, a big takeaway is engage, right?

Like, that's how you get so much out of it. Like, the speaker lineup, I've looked at it, it's amazing, but also, you know, you get out what you put into it.

And so definitely go engage with the speakers, engage with the people around you. Yeah, it's super exciting.

Emerging Trends and the Role of AI

So you've got to track about databases. There's a lot of emerging trends out there. Like, do we have to even say the word AI, right?

Like, it's a incredibly hot topic. Where do databases fit in there, right? Like, are databases dead and boring?

First year of the track, I'm super excited about it. Apparently I'm a weird nerd.

Yet what's the role of all the other things that we should be talking about? How does your track fit in with that?

How should we be thinking about broader trends and all of that, and when it comes to, you know, what's happening in the broader landscape?

Sai: Absolutely.

I think we should talk about AI because that is the emerging trend, and I strongly believe that for AI to succeed, databases are going to be very critical.

The reason I say that, right, like, I mean, AI would actually lead to demanding workloads for which databases need to live up to, right?

Like, so I'm expecting an orders of magnitude increase in requirements and demand for data.

Craig: So an "order of magnitude," are you thinking in terms of scale or user experience or both, like, 'cause I think, like, how people are working with things are different, right? So how does that impact databases?

Sai: Yeah, so orders of magnitude in respect to multiple things, right? Let it be performance, right?

AI would need databases which are 10 times faster, let it be with respect to latency, throughput, and concurrency, right?

That is one, second is, like, databases need to handle orders of magnitude data, right?

I mean, it's not going to be petabytes now, it's going to be zettabytes, right?

So that is number two, and number three, it's also going to be reliability, right?

So I see that in every dimension. Databases need to evolve to be able to support, like, AI, right?

That is the reason I think that AI cannot be successful without evolving databases. I think that's where the next set of innovation and companies, new companies probably will come up, where databases are being able to serve AI with these orders of magnitude, increase in demands and requirements.

Craig: All right, so we've talked a whole bunch about your track. I'm excited for it. Yes, it's your favorite child.

You're not supposed to have a favorite child, but your track is your favorite child.

But, like, if there's a second favorite, like, is there another track or also, and or, something else that you're incredibly excited about, in regards to the Data Council conference.

Sai: So another track that I love is going to be the data engineering track, like, right?

Like, this is probably because, like, PeerDB was a data replication tool, like, we moved data, right?

Like, but that is one track, which is, like, the root of why Data Council, like... I mean that's, like, the...

Craig: It's the origin, right? It's really the foundation.

Sai: Yeah, exactly.

Craig: And, you know, you can't get away from that and it does it so well every year, right?

Sai: Exactly, right, so that track is going to be a good one to attend.

And then there is a foundation models track as well, right, like, which goes into, like, AI, LLMs, and, like, you know, more, like, the guts of, like, LLMs, right?

Like, I think that is going to be, like, a fun one, right? Very, like, engineering and, like, technical focus, I would say.

And then there's also, like, a Lightning Talks track, right? It's going to be a lot of fun.

So last year, I gave a talk at, like, the Lightning Talks track and it was amazing. It was, like, 15 minutes of you know...

There was a talk about, like, Rust programming and, like, that impacting, like, databases, right? Like, and then I spoke about, like, you know, data movement, right?

Like, so it's, like, very, like, interesting, like, you know, quick bits across, like, the entire data space, right? Like, so I think...

Craig: I always highly recommend the lightning, I remember actually at PyCon years ago when Docker was first...

Like, Docker was first unveiled at a lightning talk at PyCon, and it's, you know, like, no spoilers, I don't know what's coming in the lightning talks.

But you never know, it's rapid fire, and the ability to convey a message in that time and you see so much so quickly that it's always a, you know, kind of a must see.

Like, any other kind of pieces or nugget, so the lightning talks, the data engineering track, the foundational model ones, like, any other kind of highlights of the conference?

Sai: Another thing that I would like to note is there are these hands-on sessions where there are speaker office hours, where audience can go and interact with speakers right away, and these office hours are right after the speaker gives the talk, right?

Like, so you can actually go spend time with the speaker, like, have a discussion, right?

So that is something that is very unique I saw in the Data Council conference where you get a chance to personally meet the speaker, have a discussion, right, like, and learn both ways, right?

So that is something that I'm very excited about as well.

Craig: Yeah, that's really, like, I... You know, usually, there's a swarm after a speaker, you know, talks and, like, it feels intimidating to go up there and you know, just even say hi or ask a question.

Like, the ability to kind of break out in a smaller format with them, that's actually...

I don't know if any other conference I can recall that's done that and that's actually really unique and actually probably takes some of the pressure off as a newbie attending a conference to go up and talk to these speakers.

Sai: Absolutely, and I think it also eases some pressure on the speaker as well because they can really hang out and talk to folks and they learn, right?

Like, because last time, I was a part of this and it was a lot of fun because, like, the talk was for 15 minutes, right, but then the hands-on, office hours went on for, like, 45 minutes, where we had so many amazing discussions.

Craig: Cool, well, I just want to say, you know, thanks so much for sitting down, Sai.

I'm super excited when you sent over kind of the rundown of the track. I was excited, I'm going to be there. You're going to be there.

Come up to us, say hi. If you're at Data Council, we'd love to see you. I'd love that databases are kind of becoming sexy again, right?

Thanks a lot to Andy Pavlo who's going to kind of kick things off for the day, you know, in the track, and then it's going to be a whirlwind tour.

It's kind of like this databases courses, all compressed in one day, one amazing track, a lot of speakers, so I'm really excited about it.

Look forward to kind of seeing folks there. I'll be there if anyone wants to talk Postgres.

Sai will be there if anyone wants to talk ClickHouse and you know, databases broadly.

We'll geek out with anyone on it. Sai, anything, you know, that you want to leave to folks?

Sai: No, that's it, very excited for the conference, right?

We started effort around six to seven months ago and the place where we are at now, right, like, with the kind of speakers.

The venue is also going to be amazing, right? Like, I mean I was speaking to Pete and he was saying that, like, it's going to be in Oakland and it's a very raw island, like, a very...

You know, it's an amazing venue so you should check that out, right? Like, so I'm sure that it's going to be a lot of fun and as Craig mentioned, right, just reach out to us, right?

We'd would love to have some great conversations, and very excited to see you there.