top of page

Future of Big Data With Apache Iceberg and Polaris

Snowflake's Ron Ortloff reveals how the data cloud company is accelerating big data analytics with Apache Iceberg and the new Polaris cloud service.

Snowflake is on a mission to enable every organization to be data-driven. With its latest innovations around Apache Iceberg and the launch of Polaris, the data cloud company is making it faster and easier than ever for developers, engineers, and architects to harness big data for transformative business insights.

Bringing Open Standards to the Data Cloud

At the core of Snowflake's strategy is embracing open standards and avoiding vendor lock-in. With the general availability of Apache Iceberg on Snowflake, customers can now enjoy the flexibility and interoperability this open table format provides.

"The whole idea with an open table format is the data is yours," says Ron Ortloff, Head of Data Lake and Iceberg at Snowflake. "When we write Iceberg, we're putting that data in the customer's own storage account. If they want to take that data and go somewhere else with it, it's theirs."

This commitment to openness is a key differentiator. "These cloud providers, Snowflake included, aren't seeing open source as a threat," Ortloff explains. "We're seeing it as an opportunity to provide customers what they've been asking for — a level playing field where we can differentiate on our platform strengths."

Simplifying Data Lake Management

A significant pain point Iceberg addresses is the complexity of data lake management. With traditional platforms, tasks like compaction and vacuum can be burdensome, often requiring manual maintenance jobs that are prone to failure.

"In the absence of what we do with Snowflake, customers would have to create these maintenance jobs on their own, schedule them, and if they fail, someone gets called in the middle of the night," says Ortloff. "We've integrated that with the Snowflake platform, so the customer doesn't have to do any of that. We take care of it all."

This automation is included in Iceberg's general availability on Snowflake, showcasing the company's focus on simplicity. "When we built and incorporated Iceberg into the Snowflake platform, from day one, we adhered to the core principles of how Snowflake is bonded and built," Ortloff notes. "It's simple to use, easy to use, and it just works."

Driving Performance at Petabyte Scale

Of course, simplicity can't come at the expense of performance, especially when dealing with massive datasets. Snowflake has rigorously optimized Iceberg throughout an extensive preview period, and hundreds of customers have provided feedback.

"We have one customer that created a petabyte table - a single table of one petabyte," Ortloff reveals. "The work we've done for Iceberg, we're at a point now where our implementation is basically on par from a performance standpoint with Snowflake's storage format."

This means customers no longer have to trade between open formats and performance. They can leverage Iceberg for its openness and interoperability while still enjoying the speed and scale Snowflake is known for.

Unlocking Analytics Across Clouds

Another critical aspect of Snowflake's approach is enabling analytics across clouds and regions. The newly launched Polaris cloud service makes this even more seamless by serving as a single entry point to an entire data ecosystem.

"Polaris has an Iceberg catalog that will federate with other catalogs," explains Ortloff. "You rarely talk to a customer with just one catalog, especially the big enterprises. They have a legacy Hive, a new Iceberg one, and maybe a department has gone rogue with their own. You want to get some semblance of understanding of where these different catalogs are, what assets exist in them, and have that single pane of glass to access it all."

By federating disparate catalogs and enabling queries across cloud storage systems, Polaris erases data silos and unifies distributed analytics assets. This empowers developers, engineers, and architects to seamlessly work with data wherever it resides, without vendor lock-in or data gravity constraints.

Accelerating Time to Insight

Ultimately, all of these innovations—Iceberg, Polaris, and the integrations with Snowflake's platform—serve one overarching goal: getting data into users' hands faster so they can drive business value.

"We have things like dynamic tables in the Snowflake platform that automates the whole silver and gold layer creation process in a lake house architecture," says Ortloff. "That declarative nature lets you distill and build data products rapidly, and soon, this will be coming into preview for external tables as well."

This emphasis on user enablement extends to Snowflake's Snowpark developer framework. Snowpark lets data engineers quickly build pipelines for cleansing and preparing data by providing a familiar Python interface. With capabilities like Snowflake's Cortex functions, previously tedious tasks have become effortless.

"Leveraging a simple Cortex summarize function to get the gist of a blob of text, that's a powerful cleansing activity that would have taken a lot of time before," Ortloff explains. "Those sorts of things will accelerate time to insight vastly."

A Bright Future for Big Data

As the world becomes increasingly data-driven, Snowflake is well-positioned to be the platform that powers the next generation of analytics. With its embrace of open standards, seamless cross-cloud capabilities, and relentless focus on performance and simplicity, the company is earning customers' trust and capturing the vanguard of the data revolution. 

For developers, engineers, and architects, this means a future where the barriers to big data analytics are dramatically reduced. Armed with tools like Iceberg and Polaris and the power of the Snowflake platform, they can focus on higher-order problems and deliver unparalleled business value.

As Ortloff puts it, "It's going to be exciting times ahead as we partner closely with customers and the whole ecosystem. Together we will keep pushing the boundaries of what's possible with data."


bottom of page