Cloud service beta and S3 support

[tarasglek]

I been using paradedb for toy projects. It's great. Would love to try S3 support and the managed service. Is there eta on either?

[philippemnoel]

What is your use case for querying S3? And what managed Postgres-as-a-Service provider are you using, if any?

[tarasglek]

We use neon. Neon charges for uncompressed data and is really inefficient with row based storage for our append-only tables.

We also use clickhouse and much prefer their pricing model for "developer" instances.

Ideally we would pay for compute and bring our own s3 for your parquet backend plus some management fee.

[tarasglek]

Our tables are append only and fairly narrow. Thus seem like a perfect candidate for s3. Clickhouse is not ideal fit for us because we like the strong postgres ecosystem, eg prisma and postgrest

We also like to append one row at a time, so some solution with incremental compaction would be cool..eg a nightly vacuum operation could move data to s3

[philippemnoel]

That's possible today with ParadeDB re: pay for compute + bring your own S3.

re: incremental compaction, this is in-progress. We are building support for S3 reads and writes, and you could do compaction to S3 on a cron job via pg_cron.

We should connect over a call and we can get you set up with everything you need here. Can you email me at [email protected]?