Snowflake, which offers a data warehouse platform as a service (PaaS), is using the cloud to address the technology bottleneck for the sharing of B2B data. Its new "Data Sharehouse" is a clever approach to helping business partners share reports and analytics, but it will require features such as managing access control and cataloging shared data to make the service scale.
Using the cloud to address the B2B data-sharing crunch
Sharing of data between business or trading partners is hardly new. The practice dates back to the implementation of Electronic Data Interchange (EDI), which trading partners in the supply chain have used for roughly half a century. EDI connections could either go direct from partner to partner or run through trading hubs that EDI service providers have operated for decades. Although those trading hubs have become internet-based (replacing private networks) over the past 20 years, setting up trading partnerships remains labor-intensive, both from technical and business process standpoints. Workarounds, such as designating trusted third-party business partners as extended staff and granting them access through internal identity and access management (IAM) solutions, can prove to be a haphazard approach.
Snowflake, which offers an analytic database/data warehouse cloud service that competes directly with Amazon Redshift, Microsoft Azure SQL Data warehouse, and Google BigQuery, has come to the conclusion that the cloud is part of the solution. It is extending its service with a "Data Sharehouse" feature that lets Snowflake customers designate specific business partners and authorize them to have access down to the database table and view level (and within the view, customers can define which columns and data ranges are visible).
The best-known (but not the only) use case for business partners sharing data is associated with B2B systems, such as EDI, where data-sharing practices have been well established for decades. But as an analytic and reporting system, Snowflake allows business or trading partners to share other information that either falls outside EDI, such as data about the possibility of weather-induced delays for shipments, or that supplements some EDI transactions, such as sharing demand forecast data with analytics. The partner, or "consumer" of data, views that data simply by paying for usage of the Snowflake cloud service; no subscription is necessary. And unlike traditional EDI trading hubs, there are no complex transactional integration "handshakes" involved.
Snowflake is on to something here, but the solution (which is generally available) is still fairly rudimentary when it comes to managing access. Today, access is managed through manual scripting at the command line; this approach will be sufficient for sharing data with a handful of partners where you can track access with a spreadsheet. Scaling it further to dozens or more business partners will require more automated solutions, providing the opportunity for third-party partnerships for managing B2B access. Likewise, the service could also be enhanced with information catalogs that provide a more qualitative background for data being exposed, such as its provenance, and providing a back channel for business partners to collaborate on fine-tuning data sharing. This is also an opportunity for partners specializing in catalogs such as Alation or Waterline.
The cloud architecture, which separates data from compute, was key to making this service feasible. By providing view-only access, it doesn't place any demand on compute, making this service potentially very scalable if third-party access management pieces are put in place.
On the Radar: Snowflake, IT0014-003029 (June 2015)
Tony Baer, Principal Analyst, Information Management