Recommendation - joshid43016/AnalyticsDataEcosystem GitHub Wiki
Differentiate analytics database platforms based on fit for pattern
While we cannot create a decision tree, we can provide recommendations based on workload characteristics. The decisioning process will have both ability to meet SLAs and best performance for cost -
-
Whether the processing is infrequent and doesn't have stringent SLAs From cost perspective, serverless platforms are better for infrequent workload which doesn't have response time SLAs. Eg: AWS Athena/RedShift spectrum / Snowflake. Both AWS Athena, AWS Redshift and Google BigQuery charges are based on data scanned while Snowflake is based on compute credits for duration when cluster is us
-
Managed platform where we have fewer knobs for indexes Fully managed platforms like AWS Athena and Snowflake are perfect cases
-
Mixed workload (tactical and analytical) where we have stringent response time SLAs Teradata IntelliCloud has a very mature workload management solution named TASM which allows you to best leverage the platform
-
Advanced SQL Teradata - Geospatial analytics
-
Security Teradata IntelliCloud allows us to use LDAP for authentication and authorization. This is currently not available with other Cloud based Data warehouse platforms
-
Data Exchange Snowflake marketplace allows clients to share data as producers and consumers thus reduce time to value, improve data quality and data freshness