Hacker News Reader

Comment by VWWHFSfQ | Hacker News Reader

I think MySQL is still significantly more widely deployed. And Postresql wouldn't be ahead of things like Firefox, apache, etc in terms of successful FLOSS. The world wouldn't really be that much different if postgres didn't exist. But it would be significantly worse if Firefox or apache never existed

View on HN

I think that is a rather naive take on the scope of PostgreSQL in "invisible" services you depend on regularly

can be swapped out for another floss database pretty easy without any real impact. internet in general would be significantly worse right now if FF or apache never took off

I wish you the best of luck replacing PostGIS with just about anything else.

Spoken by someone who has never used any db specific functionality to facilitate simplifying logic and constraints.

[laugh react]

Measuring the importance of Postgres is tricky. I’ve noticed that a lot of new DB services use Postgres tech at some level. Cockroach DB is a good example and arguably a pretty important project.

What do you mean? CockroachDB doesn't use any Postgres tech other than reusing the same wire protocol.

But that is Postgres tech and it gave CockroachDB the ability to reuse existing client code for Postgres.

https://www.cockroachlabs.com/blog/why-postgres/

I don't know why this comment is downvoted so much. Every company I've worked for uses MySQL (or a MySQL fork, eg. MariaDB or Percona). I've never seen an installation of Postgres used in production at any company I've worked for. I've been waiting for a decade to start using it, but there have always been three issues that made Postgres an unacceptable choice for my coworkers and me, at every single company I've worked for:

1. The lack of built-in replication, requiring third-party plugins based on slow triggers, etc. Postgres finally added, after many years, off-the-shelf replication, resolving this concern.

2. No equivalent to MySQL's "utf8mb4_unicode_ci" or "utf8mb4_unicode_520_ci". Postgres finally added, after many years, the "und-x-icu" collation, resolving this concern.

3. Lack of a multi-threaded model, meaning every new client connection forks a new child process, utterly decimating performance and scaling. Expecting users to throw a connection pool/proxy server in front of every Postgres installation to work around that bottleneck is unacceptable; a proper multi-threaded model needs to be provided off-the-shelf. The developers continue to fight back on this issue, and the fact is they are wrong.

If and when that final point is resolved, I might finally be able to put together a proposal showing all of the benefits and features present in Postgres compared to MySQL; and with those above three points all resolved, I might actually be able to gain traction and support with the other senior developers, and win a plea to management.

> I've never seen an installation of Postgres used in production at any company I've worked for.

many well known companies use postgresql in forward facing and internal systems. my company during 911 led the deployment of critical paging infrastructure run on postgresql.

https://postgresql.us/events/pgopen2015/schedule/speaker/74-...

Don’t your clients need their own connection pools anyways? Eg if the webserver isn’t pooling connections itself, that’s a problem regardless.

Pooler-proxies exist because some applications do not or cannot pool connections. Sometimes these are applications we just don't have the code to.

Reply on news.ycombinator.com