Approach to Data Modeling in BigQuery

Google’s BigQuery is a fully serverless offering in the cloud and is optimized for blazing fast petabyte-scale data processing using ANSI SQL alongside running machine learning and analytics processes at scale in a cost-effective manner. This white paper addresses the critical design aspects that would make your organization’s BigQuery implementation successful.

 What will you learn from this white paper?

• Challenges of data modeling in big data. Cloud data warehouses like BigQuery are designed and optimized for full table scans, whereas legacy platforms are optimized for joins and heavily rely on indexing. This difference has enormous implications as we migrate analytical data to the cloud.

• Practical tips on building a data model in BigQuery. A summary of some overarching guidelines that you want your organization to be mindful of as they start bringing more datasets onto BigQuery.  

• Design considerations. Essential tips on design consideration, normalization of data, use of arrays, fact table design, dimensions, partitioning, and clustered tables. 

BigQuery offers excellent advantages for modern data engineering needs. The approaches, principles, and critical factors discussed in this white paper will largely remain relevant to determine the performance and agility of your enterprise data and analytics, even if more features are added to BigQuery in the future.