What Does SnowFlake Do?

SnowFlake had been the biggest IPO of the year 2020 with a valuation of approximately 66 Billions. There was a lot of enthusiasm shown by the investors in the IPO but almost everyone of those enthusiast I spoke to didn’t quite understand what snowflake actually does. So in this article I try to explain as how I understand it. I will try to explain the bigger problem they are addressing, offered solutions, their moat, competition, risks to the investors and future outlook of the company.

SnowFlake is a cloud-based data-warehousing solution. A data-warehouse is a repository for big data that is being generated by your organization from different apps, device logs, databases etc. A data-warehouse efficiently manages this big data, which may be stored in several distributed databases, and extract valuable analytics and insights for the business.

Managing data at this scale has few problems. First the data has different formats,it is a mix of structured and un-structured data, querying that data is expensive operation, data sits in isolated silos and data lakes, there are governance issues, data isn’t shareable between organization.

SnowFlake offers a solution to these issues by providing a platform that consists of three layers Storage, Compute and Services. Following diagram represents these layers.

Data access and compute operations on SnowFlake utilize the cloud principles of metered access and resource scaling so you only pay for what you use and you can scale as your data and compute requirements increase. SnowFlake also provide a marketplace for data where different organizations can offer their data to the customers with specific needs.

This solution could be applied to several use cases such as creating data ware houses, creating data lakes, creating data engineering pipelines, data science projects and sharing data between organizations and these uses cases are needed in a diverse industries such as financial, technology, healthcare, education etc

Their moat is the capability to provide a solution that not only handles the management of big data but it also offers compute capabilities on it such as querying and running analytics. Other analytics solutions such as Splunk, DataDog expect data is fed to them from individual data sources or from a date-warehouse that already has that data available. Also as more and more organizations use their platform to store their data and allow it to be shared with other organizations it will become more valuable due to network effect.

I believe its a good investment because of the digital transformation that is happening around us and this transformation is resulting in Exabytes of data in a matter of days and making it difficult for organizations to efficiently store and manage this data and on top of it efficiently apply compute on the data and drive actionable insights from it. And I believe snowflake gels these two requirements in a nice package so that users can not only store their data but efficiently access it at a reasonable cost due to cloud principles of metered access without worrying about scalability.

The risks are in terms of competitions, SnowFlake uses AWS, Azure and Google cloud as the underlying cloud infrastructure and these providers have offerings that are similar to SnowFlake. Also many organization are moving towards the model of having a common data-warehouse that removes the issues of siloed data, governance and enables sharing of data for the purpose gaining broader insights for example PaloAlto networks is working towards a marketplace for security data and IBM's cloud paks. Also bigdata solutions that are already deployed by customers for their analytics needs such as Hadoop, Cassandra, Spark pose another challenge for acquisition of new customers for SnowFlake.

Overall I do feel that SnowFlake provide advantages specially considering how much data we are generating and the challenges that are associated with doing computation on data and get meaningful and actionable insights out of data. Snowflake allows organizations to not only enable them to put all their data in a query able and centralized repository rather than on disjointed and siloed data lakes but it also allows organizations to share data securely so the organizations that had limited insight on the truth due to narrow vision now have the capability to have complete truth because they can analyze data across their industry.

Related posts:

Este fue mi primer MUN del 2020 y mi primera conferencia virtual. Debido a la pandemia, se cancelaron todas las conferencias presenciales y tuve que optar por MUNs virtuales, como lo fue SOI MUN…

A brief introduction to using machine learning for default prediction at Advanon.

A cancer researcher and professor at Harvard Medical School, Sam W. Lee, PhD, has examined apoptosis’ influence on cancer formation. Sam W. Lee coauthored a paper titled X-linked Inhibitor of…