Author: Sudhendu Pandey
Introduction
Imagine you have never heard about Snowflake Data Cloud before. Or maybe you have, but never bothered in the last 8+ years to know more about it. Yes, you saw some LinkedIn posts or some Marketing billboards (aptly called Snowboards), but that is all. You are happy where you are and never felt the need to explore the product.
Suddenly it’s 2024! And you get a call from an interviewer asking if you are willing to work on Snowflake Cloud! Or you are a Data Engineer in a company and your manager knows someone who knows someone else who has worked on Snowflake and recommends it. And now your manager wants you to figure out what it is all about!
Seeing Snowflake in 2019–20 and seeing Snowflake now is drastically different! It is almost as if it is a paradigm shift. I for once, would have never imagined it would be so different!
This essay will talk about what should be your approach if you are just getting started with Snowflake. Mind you, this is not very different from what an existing Snowflake user would do, but you have the advantage of entering the ecosystem at its prime!
Not Just A Data Cloud
Firstly, we need to understand one of the crucial aspects. Its almost 2–3 years since the pivot and yet we hear it time and again:
Snowflake is just a cloud Data Warehouse.
Agreed Snowflake’s core workload which enables most of the other workloads is the data warehousing part, BUT that is not what you should be the only thing you should be looking at. Snowflake has evolved in all four directions and it could do much much more (we get to that later in the article)
Cannot crawl into flying
As a consultant, we prefer our customers not overwhelmed at the onset of their Snowflake journey. I would like them to have a good data strategy before any other strategies (AI, ML, etc.). We would like them to build the initial MVP, get and win key stakeholder confidence making the take-off smooth.
That said, it is crucial to be ready when the rubber hits the road! It is important to think through the additional use cases from the start. You may or may not be already doing it, but if you are anywhere on the data leadership chain, you must think through it.
You don’t want to miss the snack cart!
Day 01 of Knowing Snowflake
It is given Snowflake can ingest data from various sources, help you transform, model it, and get it up and read for analytical use cases. That is probably the reason why you are planning to onboard Snowflake in the first place. But that is not where you stop, Snowflake can do a lot more than just the data warehousing. Below are the representations of all workloads Snowflake supports.
Again, this might be obvious, but it surprises us how many of the customers are only using part of the product!
Things to ask for
The next section lists down potential discovery questions we ask our clients to understand eligible use cases for Snowflake. Snowflake has evolved a lot and asking only ‘data warehousing and reporting’ questions will be a gross under-utilization of the platform.
This is an ever-evolving discussion, but the idea is to explore all areas of technology and business and then do the deep-dive into individual
- Are you using any document as part of your data?
- Is there a need for near real-time data analytics?
- Are you a heavy Python user? Do you have apps created using Python that talk to your data?
- Are there any good ML use cases in your industry/domain?
- Are you purchasing/procuring any additional datasets from any provider?
- Are you planning to share your data and be a provider (monetize)?
- Do you have any use cases for LLMs? Are you using any LLMs currently?
- Are you using any containerized application that talks heavily with your data?
- Do you have techno-business users who would like to play with data apps?
- Are you dealing with Open File Formats such as Parquet, ORC, AVRO, JSON, and CSV?
- Are you dealing with Open Table Formats such as Apache Iceberg, and Delta Lake?
- Do you have any use case for geospatial analytics
Strategy Supersedes Implementation
“There will never be a better time than now” — Virginia Alison
The takeaway, for someone starting new with the platform, will be to have those outcome-based discussions right at the beginning. Once you identify the problems, you move toward the use cases and solutions that can help you solve them. Then it will be a validation exercise if Snowflake, with its workload, fits which of those use cases.
