I'm sort of afraid to ask but amidst all the ai blog spam I can't find good information. what's da deal with yno the different #database lib? I understand the difference between dbs with fundamentally different architectures like mongodb and couchdb, but like what's the deal with postgres vs MySQL vs mariadb and whatnot, and how come ppl seem to love to hate postgres?
Discussion
nostr:npub14d70xk632yuqshz7hdrnnj79j3yufrphy4u7ryekmpr7vztwvf5q8zdm4s I've run MySQL way too much and would avoid it if at all possible.
I've even slept on the couch of one of the former MySQL creators, Brian Aker. No beef with him, though I think once Sun was sold to Oracle, most of the original people who contributed to it kind of left the building?
I've had employers pay for Percona's extra mojo for MySQL and found it too, lacking in many ways that would really make this weary burnt out sysadmin's life easier.
Postgres and SQLite don't seem to get enough love IMHO, but sometimes (at least for me) such decisions were made long ago and migration to another SQL implementation is not in the cards.
Absolutely avoid at all costs: MS-SQL. Every place where MSDE is used, IMHO would be better served by SQLite too. Albeit, MS has done worse, everything with JET (which includes Exchange) is fundamentally horrific.
Sybase I thankfully haven't touched in prod in decades, but it was IMHO way too popular for way too long, perhaps because NeXT bundled it once upon a time or something inane like that?
For those who can forgo SQL, LMDB isn't exactly the "new hawtness" anymore, but is certainly way better than venerable BDB, and even cooler: LumoSQL is basically breathing new life into the SQLightning project (which was SQLite with LMDB) so I am keen to see how that shapes up over time, but when I last looked at it, it was very much a work in progress and not yet copasetic on any BSDs.
nostr:npub14d70xk632yuqshz7hdrnnj79j3yufrphy4u7ryekmpr7vztwvf5q8zdm4s I'm no expert really, but during my PhD after the 10th time rewriting my Matlab analysis code to rearrange my data in a slightly different way for a new analysis question, I made the slightly peculiar choice to spin up a MySQL database (might have been MariaDB actually, this was not too long after it forked off of MySQL) to solve my data management problem once and for all. It was using a sledgehammer to drive a nail, but by golly that nail got driven. In retrospect I could and should have solved my problem much more simply using SQLite, which doesn't require a centralized database server and lets you easily pass around your dataset in one (potentially massive) file, which is better for scientific data sharing. So I don't know about the differences between major RDBMS software, but my experience taught me that relational databases (whether implemented with RDBMS or more ad hoc methods) are amazing and seriously underused in science, and SQLite is a great tool in particular.