GCP Data Proxy

[mgorsk1]

Signed-off-by: mgorsk1 [email protected]

[feng-tao]

Does ING plan to use GCP data catalog + atlas together?

[feng-tao]

ping @mgorsk1 , the RFC makes sense though I don't have much context from GCP catalog

[mgorsk1]

We've made a cloud PoC some time ago while prepping to cloud migration. Part of this PoC was to test DC instead of Atlas as Amundsen Proxy and we've developed basic version of proxies covering basic functionalities for tables and dashboards. The ruling on what our approach we take is still pending, there are some concerns re data security in cloud but I think for metadata it should be fine.

In general DC enables extensible type system so any custom Amundsen entity can be modeled there.

As for using DC with Atlas - it's possible to sync metadata from Atlas to DC by using one of official connectors: https://github.com/GoogleCloudPlatform/datacatalog-connectors and work well (we've tested with tableau, hive, postgresql and mysqk). They cover syncing metadata from 13 db and 2 dashboarding systems.

[MrwanBaghdad]

+1 for integration with GCP and skipping the data builder. However, one concern is how to transform and enrich the table's metadata. One use case that we don't want to expose some tables i.e (depending on policy tags, columns, naming patterns..etc) in other words the level of control we already have using the databuilder.