High-Performance SQL Cluster Engine for Scalable Data Virtualization

Ugh, it feels so wrong to read a 10+ MB .patch on github, instead of being a fork + a commit.

Using FDW to scale out data is a pretty popular approach, though it is hard to get a sense of the patch, this one looks a lot like a sharded postgres impl, reminds me of Redshift or Yahoo's Everest.

Actually, we should discuss how to contribution about the source code of PGSpider.

Regarding PGSpider, please refer the following slide.

https://www.slideshare.net/Equnix/pgconfasia-2019-pgspider-h...

Has anyone had any production experience using Foreign Data Wrappers to combine the performance of different database types into one facade? IE is it actually reasonable (and something people do) to hit bigquery for a large aggregation, Neo4j for a graph query, and Redis for k/v lookup as part of a single query fronted by postgres?

I haven't used FDW directly but I would suspect the performance would largely be related to how you can get pushdown and native usage within the source system. Combining data almost must be done on the master node.

It sounds to be a PrestoSQL-like solution based on the Postgresql FDW.

Please feedback using PGSpider.

Is this similar to GreenPlum?