Snowflake's cloud-based data-warehousing service has long been associated with Amazon Web Services (AWS), which has served the company well as it looked to establish a foothold in the market. But as cloud adoption grows among enterprises, Snowflake can't stand pat with a single vendor, so the company recently made available its data-warehousing service on Microsoft Azure. The move not only provides customer choice but also enables Snowflake to enhance other capabilities, such as replication. Investors continue to show confidence in Snowflake, which is reflected in the company landing $263m earlier this year.

The 451 Take

Snowflake is building its reputation on being a native, cloud-only data-warehousing service, which up until now, was available only on AWS. AWS served Snowflake well, but the company needed to expand to other cloud platforms to truly find broader adoption, which makes the company's recent availability on Microsoft Azure all the more relevant. While providing cloud choice gives Snowflake greater market penetration, the multi-cloud strategy provides other benefits as well, such as enabling the company's replication functionality, giving customers the option to set up a disaster-recovery strategy, for instance, to other regions or to a different cloud platform. Our belief is that Snowflake does not compete directly with every data-warehousing vendor, mostly the cloud platform providers, because its service is not available on-premises. But Snowflake's profile is rising, and the company is backed by deep investor pockets, perhaps leading up to a possible IPO.


Based in San Mateo, California, and founded in mid-2012, Snowflake rolled out its first product in 2015. The company's core offering is a data-warehousing service that was initially available on AWS. While there are other cloud-based data-warehousing services available, Snowflake posits that its strategy from inception was to build a cloud-native data-warehousing service that would address some of the long-standing challenges that have arisen in the data-warehousing market, namely scaling and concurrency, for instance.

Since its inception, the company has garnered a good deal of attention, particularly within the investor community. In January, Snowflake secured an additional $263m in funding, led by Iconiq Capital, Altimeter Capital and Sequoia Capital, while also including the company's earlier investors. To date, Snowflake's total investments come to $473m, with the company claiming a valuation of $1.5bn.

While the company was tight-lipped about revenue and specific growth numbers, management did disclose that total paying customers are approaching 1,500 and includes what the company calls its 30-day customers that number about 800. What the company has learned is that many customers initially sign on for 30 days, for instance, which serves as a stepping stone to engage with Snowflake on a longer-term basis. Moreover, management notes that the workloads for those customers that have not yet entered into a proof of concept and those with longer-term contracts are relatively the same and include a mix of short-run scenarios as well as many customers running longer, more consistent workloads.

For its latest news, Snowflake is making its service available on Microsoft Azure starting in July as a public preview with general availability expected sometime in Q3 2018. The new service on Azure will be functionally similar to what has been available on AWS, which is the cloud platform the company started with. The addition of Azure is noteworthy for a couple of reasons. One is that it affords Snowflake's customers a choice that could come into play should the customer have conflicts with AWS; for instance, from a retail perspective, need a multi-cloud strategy or have existing applications already running on Azure.

The other reason, and this relates to the company's second announcement on replication, is that now with multiple cloud platforms, Snowflake can leverage it to enhance existing functionality. The company sees its replication functionality being used in a number of scenarios, such as to enable disaster recovery so that data can be replicated from one region to another within a cloud platform environment or to a different cloud platform. Another is for data sharing to any other Snowflake account in a particular region. What this means is that a customer could effectively replicate to a different region and then share data from that region with regional customers, for instance. Replication can also enable account migrations so that a customer can migrate from one region to another and/or to a different cloud platform.



Segmenting Snowflake's competitive field can be somewhat nuanced. Because Snowflake is purely cloud-based, it competes primarily with other cloud-based data-warehousing services, which can put the company in so-called 'frenemy' territory. That being the case, Snowflake would compete most directly with Amazon Redshift, Microsoft SQL Data Warehouse and Google BigQuery.

However, there are other cloud platforms, such as from IBM and Oracle, both of which provide data-warehousing services. IBM provides Db2 on Cloud while Oracle offers its Autonomous Data Warehouse Cloud. Both also provide on-premises deployments including IBM for Db2 and Oracle with Exadata, as well as the Oracle Exadata Cloud at Customer.

Still, there are other cloud-based data-warehousing services available, with the caveat that these companies do not necessarily provide public cloud platforms, although some do provide managed cloud offerings. For instance, Teradata, which promotes its Teradata Everywhere strategy, offers support for on-premises, private clouds, public clouds and the company's managed cloud environment. Pivotal is another with its Greenplum product that can be deployed on multiple clouds as well as in a private scenario. Other data-warehousing vendors include SAP HANA, Actian's Vector and Vector in Hadoop, EXASOL, Micro Focus' Vertica, Kognitio and MemSQL – all available to be deployed on public clouds or on-premises.

The distributed data-processing vendors (Hadoop) are getting in the game as well and drive a type of decoupled architecture approach, in that querying, compute and storage are separate open source or proprietary products but also include some optimizations to ensure certain efficiencies when deployed. These systems can be deployed for data-warehouse workloads. For instance, there is the Cloudera Analytic DB offering, which is also available as a beta service on the company's PaaS platform, Altus. Hortonworks offers what it calls its Enterprise Data Warehouse and MapR offers its Data Warehouse Optimization and Analytics as part of its portfolio.

James Curtis
Senior Analyst, Data, AI & Analytics

James Curtis is a Senior Analyst for the Data, AI & Analytics Channel at 451 Research. He has had experience covering the BI reporting and analytics sector and currently covers Hadoop, NoSQL and related analytic and operational database technologies.

Jeremy Korn
Research Associate

Jeremy Korn is a Research Associate at 451 Research. He graduated from Brown University with a BA in Biology and East Asian Studies and received a MA in East Asian Studies from Harvard University, where he employed quantitative and qualitative methodologies to study the Chinese film industry.

Aaron Sherrill
Senior Analyst

Aaron Sherrill is a Senior Analyst for 451 Research covering emerging trends, innovation and disruption in the Managed Services and Managed Security Services sectors. Aaron has 20+ years of experience across several industries including serving in IT management for the Federal Bureau of Investigation.

Want to read more? Request a trial now.