Stealth startup Snowflake is clandestine no longer, throwing off the wraps Tuesday morning with the announcement of a $26 million series B funding round and the rollout of a flagship product, which it’s calling the “elastic data warehouse.”
The basic idea is to do for the data warehouse what Salesforce did for CRM – transforming the product from a piece of infrastructure that has to be maintained by IT into a service operated entirely by the provider. Snowflake is built on top of Amazon EC2, but customers will never need to deal with Amazon directly.
+ALSO ON NETWORK WORLD: Microsoft (hearts) Linux, for Azure's sake +
Today’s announcement detailed the elastic data warehouse product, but didn’t provide information on exact pricing or dates for general availability. A free trial signup is available on the company’s website.
Snowflake was originally founded in 2012 by Oracle veterans Benoit Dageville and Thierry Cruanes, as well as Marcin Zukowksi, the former academic and co-founder of database software company Vectorwise. Zukowski’s Ph.D research into what’s now regarded as the key technology to improving database performance – vectorization – is the basis for Snowflake’s core offering.
The CEO, however, is a more recent arrival – Microsoft and Juniper veteran Bob Muglia was brought on board four months ago to lead Snowflake into the public eye.
Muglia said that there is massive growth in the amount of data generated by applications used to track things like transactional information.
“The transactional information tells you what happened, but the machine-generated information might tell you how it happened,” he said. “So you buy something on a website, but before you bought, you might have looked at five or 10 different products – that’s interesting information.”
Hence Snowflake, which is designed to manage, entirely out of both sight and mind from the IT department, all the complicated tasks involved in this type of analysis.
Vice president of product and marketing Jon Bock said that Snowflake, essentially, is the Uber philosophy as applied to the data warehouse.
“Owning a car, you do everything – you have to, sometimes, even change the oil yourself … even a leased car, you still have to drive it, make sure you don’t crash it, monitor all these gauges and warning lights to make sure things are working well,” he said. “But something like Uber, you basically just say ‘I want to go from here to here, and I want to go at this time,’ and a car shows up and gets you there. You don’t even think about anything beyond that.”
Bock continued the automotive analogy when talking about what he referred to as Snowflake’s multidimensional approach to scaling. Many data warehouses that scale in regard to both CPU and bandwidth/storage use, he said, don’t let those variables scale independently – so workloads that require a lot of CPU horsepower might necessitate the purchase of a higher-tier service that includes unnecessary amounts of bandwidth or storage.
“It’s kind of like if you went to a car rental place and they said ‘well, we have a Ford Focus for you.’ And then you said ‘I really want something with a bigger engine,’ and they said, ‘OK, then we can give you this massive tractor-trailer.’”