whyframeshot - stock.adobe.com

Dremio names former Splunk executive new CEO

The former Splunk executive takes over as the data lakehouse vendor's leader, aiming to raise the company's profile to demonstrate its capabilities and compete for market share.

Dremio has named Sendur Sellakumar its new CEO five months after Billy Bosworth stepped down.

Bosworth, who first joined Dremio in August 2019 as a member of the company's board of directors, left his CEO role in February 2023 for what a Dremio spokesperson termed "personal reasons."

At the time that Bosworth left Dremio, he took a job as a managing director with Vista Equity Partners in Denver, but remained part of Dremio's executive board through May.

Edward Sharp, Dremio's CFO, served as interim CEO between Bosworth's resignation and Sellakumar's appointment on July 5.

Sellakumar joins Dremio after an 18-month stint as senior vice president of commercial and construction at ServiceTitan.

Before that, he spent nine years at Splunk, a software vendor whose tools search, monitor and analyze machine-generated data. During his time at Splunk, Sellakumar served as chief cloud officer, chief product officer, and senior vice president and general manager of cloud, among other roles.

Prior to Splunk, Sellakumar spent nearly eight years in finance at Morgan Stanley and Credit Suisse.

While his appointment as CEO of Dremio represents a move up for Sellakumar, his combination of technology and finance experience has prepared him for his new role, according to Doug Henschen, an analyst at Constellation Research.

It's a good step up for Sellakumar as it's his first CEO role, and Dremio is getting somebody with a long track record at large tech companies -- including key roles at Splunk -- and in investment banking.
Doug HenschenAnalyst, Constellation Research

"It's a good step up for Sellakumar as it's his first CEO role, and Dremio is getting somebody with a long track record at large tech companies -- including key roles at Splunk -- and in investment banking," he said.

Vision for the future

Founded in 2015 and based in Santa Clara, Calif., Dremio is a data lakehouse vendor whose platform combines the structured data management capabilities of data warehouses with the unstructured data management capabilities of data lakes to enable enterprises to blend all their data together for a more comprehensive view of operations.

The vendor has raised $410 million in venture capital funding to date, including $160 million in January 2022 and $135 million in January 2021.

Recently, Dremio launched its first tool featuring generative AI capabilities, a text-to-SQL translator that enables customers to type queries and commands in natural language, which it automatically converts to code that the vendor's platform can understand. Dremio also unveiled two more generative AI-powered tools -- Autonomous Semantic Layer and Vector Lakehouse -- that are now in preview.

Databricks and Snowflake are among Dremio's competitors. But the comparison is not straightforward, Henschen noted.

Dremio's platform is based on a decentralized rather than centralized approach to data management, much like that of Starburst and Denodo. So although Databricks is also a data lakehouse vendor, and Snowflake provides a data cloud, data mesh and data fabric vendors are Dremio's closest competitors.

In addition, while Snowflake focused on embracing SQL and enabling in-database queries, and Databricks benefited from its use of Spark to fuel machine learning workloads, Dremio helped build up Apache Arrow as an open source technology for developing analytics applications. Dremio is now an early adopter of Apache Iceberg, which is becoming fundamental to data storage.

As Dremio moves forward under Sellakumar, the vendor has the potential to gain market share if it's able to show that its decentralized architecture can result in cost savings by reducing the need to copy or move data, Henschen said.

"Dremio is in the lakehouse market, but its strengths have been its distributed, open architecture focused on accessing data where it lives for BI/analytics without moving or copying," he said. "Lots of vendors are now adding [similar] query capabilities. So Dremio will have to make its case that it can match or beat rivals on performance but at a lower cost."

Another of the vendor's strengths is an open architecture that enables customers to avoid vendor lock-in, Henschen continued.

Sellakumar, meanwhile, said open architecture built on Apache Iceberg and an emphasis on optimizing for BI/analytics workloads will remain a focus under his stewardship.

"You'll see us continue to invest in Iceberg and make that a tier 1 model within our business," he said. "That's a big foundational piece. A second piece is that we've been optimizing for BI workloads -- machine learning is important, but BI workloads are the majority of queries -- and we think we can upend the economic model on the BI side."

Another focus for Dremio will be the concept of data as code, Sellakumar continued. Data as code is an approach that enables users to process, manage, consume and share data the same way code is shared during the development process.

Its aim is to increase collaboration while reducing the number of copies of data needed to collaborate by making the data available in a communal development environment. The result is cost savings.

Meanwhile, as organizations begin to develop generative AI models trained with their own data in addition to the data contained in public large language models such as ChatGPT and Google Bard, Sellakumar predicts that data as code will become more common.

"Data scientists and data engineers will find [data as code] a more seamless way to run their models," he said.

Beyond emphasizing product development strategies already in place, including completing the generative AI tools now in preview, Sellakumar said cost optimization -- particularly related to BI -- will be a highlight of Dremio's roadmap.

Lack of visibility

While Dremio offers a strong set of capabilities that are in line with where data management and analytics appear to be headed, its platform isn't known nearly as well as those of some of its competitors.

Though Dremio hasn't taken exactly the same approach to data management and analytics as Snowflake and Databricks, and the three vendors only recently have become more direct competitors, the three have occupied a similar market since Dremio emerged from stealth.

Yet Snowflake, founded just three years before Dremio, in 2020 executed the largest tech IPO in history, and Databricks, founded two years before Dremio, has raised $3.5 billion in funding.

So what Dremio needs to change is its limited visibility, Sellakumar noted. If Dremio can become more well known, there's potential for substantial growth.

"Not enough people know about us," Sellakumar said. "Our biggest weakness is [lack of visibility]. We're running some massive query volumes for very large customers, so we know we've been proven in that area. Our opportunity for growth is getting visibility to a lot more of the market."

Henschen likewise said that a better go-to-market strategy would benefit Dremio now that Sellakumar is the vendor's CEO.

He noted that Dremio appeals to customers that want to keep their data management and analytics operations simple, open and low cost, while also emphasizing data engineering. Its focus on enabling BI/analytics is appealing, as is its open architecture that allows data scientists to use the tools of their choice to work with data in Dremio.

Databricks and Snowflake, meanwhile, are expanding beyond their roots and trying to become do-it-all platforms, according to Henschen.

"I don't see [Dremio's] offering changing much, but go-to-market execution can always be improved," he said. "That will be Sellakumar's challenge, along with [potentially] taking the company public."

Eric Avidon is a senior news writer for TechTarget Editorial and a journalist with more than 25 years of experience. He covers analytics and data management.

Dig Deeper on Data warehousing

Business Analytics
Content Management