Don’t look for Snowflake Inc. to tailor its cloud infrastructure for use in on-premises environments anytime soon.

IDC has predicted that 2021 will see over 90% of global enterprises relying on a mix of on-premises, public and private clouds to meet infrastructure needs. The prominent investment firm of Andreessen Horowitz has publicly stated that the current cost of cloud may drive enterprises back into the privately run data center world.

Yet cloud-native Snowflake, which reported triple-digit revenue growth year-over-year in its most recent quarterly earnings, will not be enticed into delivering tools for a hybrid model, according to one of its founders.

“As of now, I don’t see any reason for us to go on-premises,” said Snowflake’s Benoit Dageville (pictured), co-founder and president of products. “Snowflake is growing like crazy; the world is moving to the cloud. I would say 90% to 99% of the world is moving to the cloud, maybe 1% is coming back for some very specific reasons. In the end, we might miss a small percentage of the workloads that will stay on-premises, and that’s OK.”

Dageville spoke with Dave Vellante, host of theCUBE, SiliconANGLE Media’s livestreaming studio, at AWS re:Invent. They discussed the rationale behind the firm’s cloud native vision, how the company filled an enterprise need for managing large datasets, expanding interest in a data mesh and Snowflake’s refusal to become embroiled in a “benchmark war.” (* Disclosure below.)

Elasticity in the cloud

Dageville’s comments about remaining cloud-native were consistent with remarks made by Snowflake Chief Executive Frank Slootman in an interview with theCUBE one year ago.

“We’re not doing this endless hedging that people have done for 20 years, sort of keeping a leg in both worlds,” Slootman said. “Forget it. This will only work in the public cloud. People will come to the public cloud a lot sooner than we’ll ever come to the private cloud.”

The company’s original vision was to build a service that could scale for workloads and be able to query a huge volume of data. The cloud offered a viable way to may this happen.

“The cloud had elasticity; you can run every workload that you want concurrently in parallel on different computer resources, and you can run them against the same data,” Dageville said. “All of a sudden our customers don’t need to manage infrastructure, they don’t need to patch, they don’t need to tune. Everything is done by Snowflake the service, and they can just load in and run their query.”

What also propelled Snowflake toward success was enterprise frustration with existing frameworks for distributed processing of large datasets, such as the open-source Hadoop.

“Everyone knew Hadoop was really bad, complex to manage, slow,” Dageville said. “It had good aspects. This was the only system that could manage petabyte scale datasets, and cheaply, which was good. We wanted to do that plus have all of the good attributes of a data warehouse system.”

Forging the data mesh

Over the past 11 years, Snowflake has refined its model to include a seamless process for connecting organizations and analytics teams to relevant data whenever it is needed. This incorporates the principles of a data mesh, as originally envisioned by Thoughtworks Inc.’s Zhamak Dehghani.

Snowflake has built these principles, which are domain-driven ownership, self-service infrastructure and federated governance, into its own version of the data mesh.

“When you are a large organization with different independent business units, everyone wants to manage their own data,” Dageville explained. “Having it centralized in IT is not practical. At the same time, you really want to be able to connect different datasets together. That’s the data mesh architecture.”

Snowflake’s rise has not been without pushback from its competitors in the data warehousing space. In early November, Databricks Inc. published TPC-DS benchmark data showing that its SQL platform had outperformed Snowflake.

Snowflake refuted the results in a lengthy blog post written by Dageville and his co-founder Thierry Cruanes 10 days later. The two co-founders also took issue with the entire notion of making competitive performance claims based on benchmarked data.

“We’ve said from day one, we would never again participate in this really stupid benchmark war because it’s not in the interest of customers,” Dageville said. “TPC was really important at some point, and it is not really relevant now. We want to invest our engineering effort and our IP in solving real-world performance issues. It’s all about under-the-covers improving Snowflake and getting the benefit of this improvement to our customers.”

Dageville is particularly enthusiastic about Snowflake’s ability to make inroads in vertical markets. He believes that the company’s role in building a seamless web of data will make a significant difference as the world becomes even more connected over time.

“You can see the enthusiasm for Snowflake Data Cloud in the vertical industry,” Dageville said. “We’ve seen complete vertical industries latching onto that concept and collaborating via Snowflake, which was not possible before. The worldwide web of data that we are creating is going to be amazing. We are really at the beginning of this interconnected world.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of AWS re:Invent. (* Disclosure: This is an unsponsored editorial segment. However, theCUBE is a paid media partner for AWS re:Invent. AWS and other sponsors of theCUBE’s event coverage have no editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE