Nov 16, 2021 · 33m · mad
Fireside Chat: Aaron Katz (Co-Founder & CEO, ClickHouse) with Matt Turck (Partner, FirstMark)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this MAD Podcast episode hosted by Matt Turck, ClickHouse Co-Founder & CEO Aaron Katz discusses the evolution of real-time analytics, database market convergence, and ClickHouse’s key architectural differentiators. Katz details the company's corporate spin-out from Yandex, its Apache 2.0 open-source commercialization strategy, enterprise cloud roadmap, and hybrid go-to-market motion for scaling real-time analytical workloads globally.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 16.7% of the talking time here. How this is scored →
Matt as informed peer 3.9 Guest teaching 4.1 Guest disagreement 0.8 Matt pushing back 1.8
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 29:24 Aaron rejects 'be-all end-all' database framing
Aaron explicitly rejects Matt's hyperbolic framing that ClickHouse wants to be the 'data store for all things', stating clearly that he won't go on record claiming to be the be-all end-all for every workload.
Hardest push from Matt ▶ 19:29 Matt presses guest to identify product trade-offsMatt directly challenges the guest to move past hype by asking explicitly about ClickHouse's limitations, trade-offs, and areas where it is not the right choice.
Biggest teaching moment ▶ 4:32 Aaron reframes real-time analytics dynamicAaron reframes Matt's skepticism about real-time hype by providing precise technical parameters and highlighting that performance must be evaluated alongside cost efficiency.
Matt holds his own ▶ 28:34 Matt cites specific open-source competitorsMatt demonstrates deep sector knowledge by accurately naming open-source competitors like Druid and Pinot and asking for a sharp distinction between market claim and technical reality.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Mega-Trends in Databases and Real-Time Analytics | 2 | 4 | 0 | 0 | Matt opens with a broad high-level question about database mega-trends. Aaron takes over with an extensive overview of OLAP vs. OLTP convergence, NoSQL evolution, Stephen O'Grady's article on general-purpose databases, and ecosystem integrations. | |
| Defining Real-Time Analytics and System Performance | 4 | 5 | 2 | 3 | Matt brings industry context by noting how real-time analytics was historically hyped as a 'nice-to-have'. Aaron reframes the premise by clarifying that real-time depends on technical definition (millisecond latency over petabytes) and introduces the critical balance of performance versus cost efficiency. | |
| Real-World Use Cases: Historical and Observability Data at Uber | 2 | 4 | 0 | 0 | Matt prompts Aaron to elaborate on historical and predictive analytics. Aaron uses Uber's logging infrastructure as a real-world case study to explain combining real-time streaming data with historical queries, while noting predictive modeling is still on the future horizon. | |
| Origin Story: Spinning ClickHouse Out of Yandex | 2 | 3 | 0 | 0 | Matt invites Aaron to share ClickHouse's spin-out story from Yandex. Aaron provides a detailed narrative on working with Yandex leadership, recruiting key executives, securing venture capital, and convincing Moscow engineers to transition. | |
| Open Source Licensing Strategy and Cloud Provider Partnerships | 4 | 4 | 0 | 2 | Matt jumps in to steer the conversation toward open source licensing and cloud provider dynamics. Aaron outlines their Apache 2.0 strategy, contrasting it with AGPL/SSPL approaches, and explains plans for building a managed cloud service. | |
| ClickHouse Architecture and Competitive Differentiators | 5 | 4 | 0 | 3 | Matt shows strong domain context by naming exact open source competitors like Druid and Pinot, asking for the 'claim and reality' of ClickHouse's performance. Aaron breaks down the architectural advantages of columnar storage, vectorization, and CPU resource efficiency. | |
| Limitations, Non-Use Cases, and Ecosystem Integrations | 5 | 3 | 0 | 4 | Matt directly presses Aaron to identify product trade-offs and non-use cases. Aaron candidly agrees and cites hosted APM tooling like Datadog as a clear example where ClickHouse is not the right choice. | |
| Product Roadmap, Security, and Cloud Architecture | 5 | 5 | 2 | 0 | Matt leverages Aaron's past experience as Elastic CRO to frame a question around PLG versus enterprise GTM strategy. Aaron delivers an insightful overview of open source distribution, enterprise sales execution, and internationalization signals, while defending their Russian engineering team heritage. | |
| Industry Convergence and General Purpose Data Stores | 6 | 5 | 3 | 4 | Matt synthesizes the macro theme of OLTP/OLAP convergence and challenges Aaron on whether ClickHouse aims to become 'the data store for all things'. Aaron explicitly pushes back on the hyperbolic framing, clarifying that ClickHouse is part of an industry trend rather than a universal replacement for all databases. |
Statements from this episode (16)
Aaron Katz predicts convergence between OLAP and OLTP database technologies
“I think we're going to see this convergence between the two.”
Katz: Data storage is being commoditized as stack value shifts
“I think storage is quickly being commoditized and, you know, the value on the stack is both on the front end and on the back end.”
Katz: eBay ingests one billion events per minute into ClickHouse
“EBay, for example, who is ingesting literally a billion events per minute into Clickhouse and then running very complex queries against that data set.”
Katz: ClickHouse handled 30 billion daily events when built at Yandex
“When it was originally written inside Yandex, the data volumes that they were seeing were at the time, you know, some of the highest in the market, you know, let's say, thirty billion events per day.”
Katz: Uber adopted ClickHouse for logging infrastructure to lower costs
“They were evaluating a variety of technologies for their logging infrastructure for their logging platform, which is being used by hundreds of developers every day for a variety of analytical workloads. And the current technology presented a bit of a cost. Con…”
Katz: ClickHouse focuses on real-time analytics, not predictive modeling
“In terms of predictive analytics and modeling, I think that's kind of the last, I wouldn't call it the last mile, but it's something that we're thinking about in the future. It's not something that we are primarily focused on today. We're more focused on the r…”
Katz names Mike Volpe and Peter Fenton top open-source VCs
“And I believe, and there's a lot of very smart people that are deploying capital in this category, specifically open source and infrastructure, but my belief, and I'm biased obviously, is that they are the top two or, you know, two of the top five, let's say.”
ClickHouse plans to double core open-source engineering team
“We're going to double the size of Alexei's team as soon as we can”
Katz: ClickHouse had second most active GitHub contributors in 2020
“I think it was the, you know, the second second highest open source database management system on GitHub last year in terms of active contributors and perhaps the fastest growing open source database management system on GitHub alone.”
Katz: ClickHouse will keep its Apache 2.0 open-source license
“We feel that today the right decision for the community is to stay with an Apache two license, and we're excited about that. Like I said before, you know, we're not moving away from open source, quite the contrary.”
Katz: ClickHouse will build a multi-tenant cloud-managed service
“And at the same time in parallel, we are going to build A multi-tenant managed service in the cloud, which will inevitably be deployed on a variety of different cloud platforms, whether it's AWS, GCP, Azure, whether we go to China, like I've done in the past a…”
Katz recommends Datadog for hosted APM over stitching custom solutions
“If you are looking to deploy a hosted APM solution, and you asked me objectively, my recommendation, I'd say use Datadog. I'd say that the product, the maturity, the experience is, is considerably better than trying to stitch together something using ClickHous…”
ClickHouse has no plans to build its own visualization layer
“Today, we don't have plans to build our own visualization layer, like you've seen other open source companies do. We plan to partner.”
ClickHouse is relocating its engineering team to Amsterdam
“We're actually, they're in the process of relocating to Amsterdam to make it just easier for all of us to do business together.”
Katz: Elastic hit $500M+ ARR with 41% international revenue mix
“When I left Elastic, I believe the company was doing a little bit north and a half a billion in recurring revenue at the time. I'm almost certain it was north of 40%. I think 41% of revenue was coming outside of the U.S.”
Katz bought clickhouse.com before announcing spin-out to avoid price gouging
“Yeah, clickhouse.com, which to Mark's earlier point is a domain that I purchased earlier this year when engineering this spin out, I thought I better buy this domain now because it's going to get a lot more expensive when we incorporate the company and announc…”