When you have 60,000 tables and views across hundreds of schemata sitting on top of a 3 petabyte storage footprint, automation and access privileges are key if you want to expose these data through a RESTful API. Our University's research service leverages Python, Django, Django REST Framework, and PostgreSQL to accomplish this. We are continuing to open source the tools we have built to make this possible!
Databases stay relevant by continuing to reinvent themselves to serve new technologies further up the stack. The latest buzzwords further up that stack are APIs and microservices, so prevalent that it is hard to see a tech advertisement that doesn’t mention them. While related to “the cloud” and “big data”, whatever the heck those terms actually mean, APIs and microservices have a slightly less annoying marketing schtick and more concrete relations to relational databases.
But what do these relations look like in practice? In this talk, the speaker will show how his team at the University has evolved from providing financial data exclusively in SAS data formats to a robust backend powered by PostgreSQL, which allows financial research to happen in many ecosystems: still available in SAS, but also R, Python, Perl, Matlab, Julia, and more. The speaker will present a case study of using Django REST Framework to build an “API through introspection.” This case study will show how Django web site and RESTful web service were built by introspecting financial data stored in a PostgreSQL database cluster.
The models for the ORM, serializers for the RESTful API, views for presenting the data to a user, filters for refining queries, URL routing, web browsable interface, user token authorization, and permissions, are all created by introspecting the PostgreSQL database information schema, all with Python. These components have now been open-sourced, including the endpoint spreadsheet exporter and the generic automated API builder.