Third day in the snow @ Snowflake Data Cloud Summit 2024

After two days filled with countless announcements - I didn't even attempt to list them - and eager audiences, today was definitely more focused on in-depth discussions and hands-on sessions. The builders keynote was an amazing comedy, illustrating relatable scenarios in the data and tech world, showcasing how complex issues can be tackled with the right resources and strategy.

I've attended fewer sessions today, but each one was packed with important insights and lots of ideas. Quality over quantity! Overall, today was another fruitful day, leaving me enriched with a wealth of content.

Didn’t catch my previous articles on the first two days of the event?

Check them out here:

• Day 1 highlights
• Day 2 highlights

Additionally, you can find the official source listing the announced updates here.

Note that it only includes features that are at least in Public Preview (PuPr): start checking this reduced list for a pragmatic approach that focuses on the closer releases.

Here's a brief summary of what I experienced today, beginning with my key takeaways!

?? Key takeaways

While technological advancements continue to enhance platform capabilities, it's essential to remind that the platform alone cannot drive business success. The platform may become increasingly advanced and autonomous, but it's the strategic alignment that truly govern its effectiveness.

To join this buffet of tempting and promising delights it's wise to adopt a strategy that considers context, priorities and objectives, governing both implementation and long-term value creation.

Embracing a buffet-style approach may lead to accumulating many shiny yet ultimately short-lived tactical solutions. As the initial excitement fades, managing these solutions becomes increasingly problematic and costly. Agility in data management isn't just about adopting the latest tools: it's about cultivating a dynamic culture that fosters innovation and collaboration and to develop an effective and sustainable strategy to pursue the goals - with the most promising and cost-effective tools and the expertise of partners.

??The builders keynote

It was truly impressive! A theatrical performance depicted typical scenarios within a data-driven organization, addressing new business needs requiring innovative tools, agility and good team cooperation. The right touch of english humor combined with practical demonstrations showed how easily solutions can be developed using LLM through Cortex, Streamlit and RelationalAI - one of the most promising native apps for simplifying graph analytics management. I've found the format original and effective, breaking the monotony of overly frontal presentations with typically lower engagement. The audience connected with the recreated scenes, leaving with valuable reflections and.. an extra chocolate bar! - placed under each seat and announced during the show to engage the audience even more.

Today's sessions

Below ?? you will find a quick summary of the sessions that I’ve joined together with my own thoughts.

? Building open data lakes and lakehouses with iceberg

?? Ron Ortloff , Senior Product Manager, Snowflake

?? Martin Lee , Senior Product Manager, Snowflake

??The talk covered how Apache Iceberg's open storage standard empowers data democratization across diverse data lakes, enabling flexible use of different data processing engines. The session was compelling for showcasing innovative ways to enhance data interoperability and streamline data workflows.

?? the wide adoption of open file format and the increasing commitment of major players to contribute to open standards fosters interoperability and polyglotism - two key ingredients for sustainable solutions and mature echosystems. Iceberg tables allow customers to avoid lock-in and to generate multiple copies of data for each engine required while reducing costs.

? How CHS saved millions on inventory management with Snowflake and Streamlit

?? Riley Buss , Senior Manager of IT Engineering, CHS Inc.

?? Zac Zinda, Principal Data Scientist, phData

?? The demand-planning process was costly, relying on intuition rather than on statistical approaches, leaving much room for improvement. Manual intervention was needed to integrate and adjust forecasts. To modernize and reduce costs, a solution was developed using Dataiku for data preparation with pushdown on Snowflake, ML out-of-the-box for forecasting and Streamlit allowed business users to refine and adjust forecasts, blending automation with human insight. The approach brought a modern and efficient solution achieving operational efficiency and more accurate forecasts.

??The talk was very engaging, showcasing a common scenario where tactical solutions developed in Excel - because it's easy, just works and gives you full control - eventually prove unscalable, inaccurate and costly to maintain and execute. Result: a tactical solution becomes an obstacle. Shadow IT, anyone? In such cases, with the advanced resources now available and proper guidance, technological debt can be mitigated, enhancing efficiency, effectiveness, and cost savings.

? Build, Deliver and Govern Data Products at Scale Using Data Mesh Principles

?? Sean McIntyre , Senior Solutions Architect, dbt Labs

?? Doug Guthrie, Senior Solutions Architect, dbt Labs

??The hands-on lab focused on developing a solution based on dbt and Snowflake to apply data mesh paradigm. It is public and you can find the full lab here.

??Even those familiar with both products will find it interesting to follow this lab. I would like to highlight the potential to control the tags (assigned to data in Snowflake) directly within dbt models : a practical approach to avoid inconsistencies between governing tags in Snowflake and an externally managed evolving model. Enterprise edition of dbt enable model sharing across projects - a must-have to facilitate data product organization and clear ownership division.

Fun extras

?? Finally did it! Got my pic with the legendary white bear - the real reason I came all this way! Did you know the person inside has a time limit before they pull him out to avoid running out of air?

Tech and beyond

Take your time to dive into San Francisco’s best spots! Swing by Fisherman's Wharf and head to Pier 39 to see the sea lions in action. Their hilarious barks and funny faces are a must-see - way more entertaining than any panel! ????

???? Follow me if you liked this content, I weekly share my thoughts on a broad set of data-related topics!