Hacker News Reader

Comment by thecopy | Hacker News Reader

I am beginning to see a trend on HN. Teams are moving away from new and young databases to old and "boring" databases as their need for stability and correctness increases.

View on HN

Which is why we grey beards just wait for the caravan to pass until the dust settles, then we get back to action and recover our dues.

Would be interesting to hear how they arrived at the RethinkDB/Elasticsearch combination over Postgres at the start of the project. It doesn't seem they lost any features after the transition. Was something lacking in Postgres at the time of initial implementation?

However, they do say:

> leaning heavily on Haskell in order to fill in some of the gaps quickly.

So it looks like they had to do some work to cover some features from rethinkdb/elasticsearch.

The code in the post is just used to migrate from an untyped interface to a typed one. Unless I'm missing something, there seems to be no mention of any missing features in postgres.

They did mention missing JSON schema validation in Postgres.

And also many mentions of no schema validation in Rethink, so that's probably not a gap.

hmm not really, it doesn't make any sense ( text search included ) to put everything into PG.

PG doesn't even have a clustering solution out of Citus, how do you scale / HA postgres using the default setup without doing sharding yourself?

What? There are over a dozen. See:

https://wiki.postgresql.org/wiki/Clustering

Some commercial.

Those are third parties extensions, ES does clustering out of the box well. Most of the thing on your list don't have proper clustering solution built-in which means you have to support it yourself.

You said that PG had no clustering solutions, I showed you there were. Some of them may require more work than others but there ARE clustering options none the less for PG. It's inaccurate to say there are none.

Reply on news.ycombinator.com