Snowpark enables organizations to safely run AI inside containers
According to a Snowflake press release, “containers package code in any language to ensure portability and consistency across environments, especially for advanced AI/ML models and full-stack data-intensive apps. It has emerged as a modern approach to
The company announced enhancements that will enable organizations to accelerate the development and creation of workloads that previously required significant time and computing power to manage. These workloads are further complicated by the fact that they cannot run directly where the data is managed. This makes “security management of data used in any of these scenarios very difficult, and in some cases nearly impossible,” Snowflake noted.
These limitations led the company to develop Snowpark Container Services. This enables “developers to deploy, manage, and manage containerized workloads (jobs, services, and service functions) using the secure Snowflake management infrastructure with configurable hardware options such as GPUs. It will be “extensible”.
expedition: Find out what your organization needs to know about containerization.
Torsten Grabs, senior director of product management at Snowflake, said the new container service was the most exciting announcement at this year’s summit.
“This is a very fundamental feature we are releasing and it opens up so many doors for us and our customers to create more value with their data.”
“We have been talking about leveraging data for a very long time, but all these small reasons were motivating customers not to do it and create data silos elsewhere. I think those days are over,” Grubbs said.
Snowflake’s Container Service Helps Reduce Complexity
“Complexity is very detrimental to an organization because it slows everyone down, creates risk, and creates additional costs. It’s about providing a single, unified product experience,” explained Grubbs.
That philosophy also applies to Snowflake’s thinking about containerized computing.
“We want to keep this as simple and straightforward as possible and help our customers create fewer and fewer silos in their data stacks,” Grubbs said. “Also, having fewer silos has reduced the complexity of his overall architecture.”
“This single product philosophy really carries over to our way of thinking about security for containerized computing products,” he continued. “The same role-based access control that our customers have, which we apply to literally every concept, also applies to Snowpark Container Services.”
The product is designed to minimize complexity by eliminating many security policies that apply to environments with multiple platforms. “This is exactly what we want to avoid because we have a consistent security governance and privacy philosophy across the board and want to apply the same concepts everywhere,” Grubbs said.
