Game-changers from Snowflake Summit 2022
The hottest features announced at Snowflake Summit 2022 are set to shape the future of the data cloud.
Snowflake Summit 2022 was a four-day extravaganza of feature unveils, product deep-dives, and use case highlights. The CEO, Frank Slootman, called it “the most consequential conference in terms of announcements in four years.” Attendees got a first-hand look at exciting innovations such as on-premises support for external tables, Python support via Snowpark, native Iceberg tables, and more.
Every announcement generated considerable buzz as talk turned to use cases and applying the enhancements to real-world challenges, but two stood out in particular as complete game-changers. They were:
- Hybrid Unistore Tables
- Streamlit integration for native data applications
Let’s explore them in detail to understand what they have to offer.
Hybrid Tables
Until now, tables in Snowflake’s Data Cloud have been optimized for OLAP (online analytical processing.) An OLAP system is designed to quickly process large amounts of data across multiple dimensions. It’s a natural fit for data warehousing and business intelligence, where data needs to be aggregated and analyzed at scale.
However, this meant that OLTP (online transactional processing) use cases that handle large volumes of transactional data — like bookings and online purchases — had to be handled elsewhere. Until now.
As the name suggests, Hybrid Unistore Tables offer the best of both worlds: a single, unified data set to develop and deploy applications and analyze transactional and analytical data together in near-real-time.
Hybrid tables will require primary keys, enforcing a unique index, and supporting additional non-unique indexes to accelerate analytics. Hybrid tables will also allow row-level locking and other features to support building and running enterprise transactional apps directly in the data cloud (more on this below).
This opens Snowflake up to a whole new set of users who have previously had to rely on other database vendors for transactional and operational needs. And, since operational and analytical workloads can now be run on a single platform, the need to run ETL jobs to move and synchronize data across multiple systems.
There’s much to like with Hybrid Unistore and the possibilities these tables offer. Data teams that have until now neglected to pay attention to primary keys and other database constraints can take the time to brush up on the fundamentals of dimensional modeling while they wait for this feature to come out of private preview.
Native data applications using Streamlit
Although Snowflake’s $800M acquisition of Streamlit was announced in March of ’22, Summit attendees got to finally see the integration for themselves. The integration was a natural fit for the two companies, given their shared goal:
“We have both the same vision — Streamlit and Snowflake — which is all about democratizing access to data.”
Benoît Dageville, co-founder and president of products at Snowflake
Streamlit allows Snowflake users to mobilize their data by embedding it in an application using a few lines of code. An app allows customers to put data into the hands of users beyond the database and BI tools. With Streamlit, developers can focus on the app’s core function (e.g., ML predictions, budget planning, forecasts) while bootstrapping the UI.
This three-minute video shows us how simple the entire process can be:
This functionality is exciting in any context, but, coupled with the announcement of Hybrid tables, it opens the doors for even greater use cases that require row-level speed and precision.
Data Marketplace is now: Marketplace!
Sensing the disruptive potential of what Streamlit brings to the table, Snowflake has rebranded its Data Marketplace to simply “Marketplace” to allow apps built directly on the platform to be shared and monetized.
What Snowflake once did for data democratization with the Data Marketplace, it now plans to do for apps. This is a thrilling prospect for anyone with a great app idea. Like the iTunes or Google Play Store, Snowflake Marketplace will make finding and integrating data apps a smooth and enjoyable experience.
Wrapping up
Boy, did Frank deliver on his promise of “consequential” feature announcements. Speaking with attendees on the conference floor and fellow Data Superheroes, it was hard to miss the enthusiasm as minds bubbled with exciting new possibilities.
“Disruption” is a word that gets thrown around all too lightly nowadays. Still, it’s hard to deny that Snowflake has opened up new avenues to previously-untapped markets and use cases with the features they announced. Benoît was especially convinced on this point:
“First, we disrupted data analytics, then we disrupted collaboration. Now, we’re about to disrupt data application development and product distribution.“
Benoît Dageville, co-founder and president of products at Snowflake
Hybrid tables and Streamlit made the biggest impact on me, but there were many more announcements that I have yet to think through. For a summarized list of everything announced at Snowflake Summit 2022, I recommend this great recap from fellow Data Superhero Frank Bell.
These features should keep me busy for a while. At least until the next Snowflake Summit, which, incidentally, has also been announced: June 26 -29, 2023.
