Materialize CEO Arjun Narayan on Ubiquitous Streaming Databases and Building User and Peer Communities
Editor’s note: This interview is part of Mission North’s Boundary Breakers series, an ongoing exchange with executives and industry observers on the business impact of new tech.
Arjun Narayan discovered his passion for databases while studying for a Ph.D. in distributed systems and big data. The co-founder and CEO of real-time streaming database company Materialize grew increasingly aware of the immense depth and promise of that technology area.
“Databases are one of those technologies where 90 percent of it is below the water level and, until you prod, poke and look, it’s not necessarily obvious what’s there,” Arjun said. “It appealed to me to work in a field in which an incredible amount of thought had already been invested.” He worked first as a software engineer at Cockroach Labs before co-founding Materialize in 2019.
What keeps Arjun fascinated by databases is how they enable users across organizations and industries to gain value from their data often by investing only a small amount of effort.
“A lot of folks are using extremely sophisticated databases even when they don’t necessarily know they are,” he said. “They may be pointing and clicking on a BI tool or a dashboard which is firing extremely complicated SQL queries under the hood that are getting crunched across a modern cloud database.”
I talked recently with Arjun about the current flowering of all types of databases from streaming to graph to vector, as well as his thoughts on the importance of community-building for startups. What follows is an edited version of that conversation.
How would you characterize today’s plethora of new databases?
We’re in a specific wave of changes in database technology enabled by the cloud where many more types of databases and vendors are emerging. You have to understand the current wave that you are in so that you’re swimming with the ecosystem as opposed to against it, which can be very painful.
Back when databases were on-premise software that you purchased and ran yourself using specialized DBAs (database administrators), there was a limit to the number of technologies and vendors any given organization could support. With the cloud, if tools work well together as good ecosystem players, organizations can adopt a larger set of tools that are fit-for-purpose and the best tool for specific jobs.
How do you see databases evolving in terms of expanding functionality and usage?
I’d classify it in multiple dimensions, with the first dimension being increasing amounts of data. Every transaction that used to be a single entry at the point of sale is turning into a lot more data points. For instance, data points that capture all the customer interactions we had leading up to the sale and/or during the sale, as well as the things that the customer tried that didn’t work out.
The second dimension is about getting much more value out of the data we already have. Once you collect more data, you can do more things with that data, which then causes you to derive more value from it, which leads you to want to collect even more data, and so on.
“A great way to think about the journey towards streaming data is making more use out of the data that you already have.”
What are your favorite examples of why and how companies are using streaming data?
A great way to think about the journey towards streaming data is making more use out of the data that you already have. You’re doing more data science and data analytics, and you are operationalizing and automating some of those observed signals and actions.
Once you start to automate these actions, you gain a lot more value the faster you respond to the signals. One obvious example is fraud detection in credit cards—it’s much better to catch it now than tonight when we’re going to clear the transactions. Sales and marketing is another big area, as is logistics and supply chain. There’s no lower boundary on how quickly you can flag that inventory has run out. The best time was literally as soon as you knew inventory was running out, the second-best time is now.
People who’ve adopted streaming data want to move as close to instantaneous as they can without throwing out everything they have and keeping their business logic as-is. They are analytics-first organizations and are looking to pull their human analysts out of the loop by moving the taking of actions on their insights from a manual to an automated process and are then looking to cut that time down.
“People who’ve adopted streaming data want to move as close to instantaneous as they can without throwing out everything they have and keeping their business logic as-is.”
Why is it beneficial for a startup to build a user community early on in its development?
The single most important thing is to communicate with people, whether they are potential users, customers, advocates, detractors or competitors. There are outsized returns in being as brutally honest as possible. Developers want to hear the ground truth, and the clearer you can be, and not oversell or overstate your case, the more they are willing to trust you.
When trying to solve hard problems, startups may feel pressure to sugarcoat what they’ve achieved because otherwise it seems so daunting to acquire users. Databases are so large and capital-intensive that there are a lot of folks, over the years, who have come out the gate saying, ‘We’ve solved it all, you can trust us.’
Essentially, they think they have to sound as similar as possible to Oracle, otherwise why would anyone choose them over Oracle? I think that reasoning is flawed, because the way to gain trust is first to be honest about how hard these problems are, and then to show users how you’re going to solve them. That approach builds more trust in a community than anything else.
“Developers want to hear the ground truth, and the clearer you can be, and not oversell or overstate your case, the more they are willing to trust you.”
What about the importance of establishing a community with other startups?
A community of advisors and other peers, potentially competitors, is very valuable because there’s a lot of nuance and counterintuitive things that you have to do as a startup. In general, it boils down to the maxim that you need to have a really good preview of the challenges your company will face six months from now, which are not always clear today. If folks can alert you now to what may break down later, you can start addressing those issues right away.
It’s a little bit like a relay race, where the next person starts running well before the baton gets to them. There’s a difference in performance between an organization that has advisors who tell them, ‘Hey, start running now, it will take you a while to build up the speed’ versus an organization where you’re starting from a standstill every time, essentially doubling the time it takes to achieve anything. In so many startups, time is in extremely short supply.