Ever since 7.1, I'm a near-fanatic PostgreSQL-user, if not to say fanboi.
I love, love, love PostgreSQL for its performance, its features, its awesome community and just for the fact that it's getting better and better every year.
Congratulations to the PostgreSQL team.
I've said it before and I'll probably say it many more times: You guys rule.
It really is amazing how consistently better it gets. The community is ridiculous, too. Every time I read through or ask questions in #postgresql I am reminded that despite knowing way more SQL than your average developer, I am still a big noob compared to the PG folks.
Most SQL questions I field from fellow devs go like:
Him: How do I write a query to do X?
Me: Which RDMS do you use?
H: MySQL
M: Well, I don't use that, but you can probably use Y
M: Now, have you ever considered switching to PostgreSQL...
SQL is a standard with approximately zero fully-compliant implementations. What's more, every database has a variety of features that aren't covered by the SQL spec, or use different syntax for some feature for historical reasons.
I've got a lot of respect for the PostgreSQL devs for occasionally breaking backwards-compatibility in the name of ANSI-SQL compliance, and for clearly marking deviations from ANSI-SQL in their reference documentation.
Since you are someone who is seriously into postgres, can I ask you, how do you deal with the lack of a serious solution for replication / hot standby? Why is this not a more pressing issue for the postgresql developers?
I ask this in the following context: I frequently argue for postgresql but in a majority of cases it gets shot down because MySQL is perceived to have a better solution for maintaining hot standby and people actually care about availability more than all the other retarded issues MySQL has put together. Unbelievably, after losing the publicity war with MySQL, postgres actually seems to now be proceeding to lose the technical war on one single issue ... yet I don't see much urgency in the postgresql camp to care about this.
how do you deal with the lack of a serious solution for replication / hot standby? Why is this not a more pressing issue for the postgresql developers?
I went to the SF Postgres meetup a month or two ago where they talked about new features in 8.4. Josh Berkus, one of the core devs hosts the meetup.
Anyhoo, hot standby _is_ a huge issue for them. We talked about it for 20 minutes. They had a dev who worked his tail off on the issue for several months. There was a lot of hope that hot standby was going to make it into the 8.4 release, but there were a few too many bugs on corner cases for them to ship the feature in 8.4 without a big refactor.
Postgres core is very concerned about replication, but shipping a stable product is their biggest concern. So, good news is that hot standby is a big priority for 8.5. Bad news, is the release isn't slated until next year.
Londiste from skytools (backed by skype) is an excellent replication product that matches mysql replication. We use it, it's bulletproof and dead easy to administer.
However the lack of hot standby is definitely a keenly felt issue...it can take an entire day to warm up a new slave.
MySQL replication is pretty broken too. It was one of the biggest problem ares while I worked their, tons of customer issues with it, slaves constantly getting out of sync with masters. I think it's fundamentally because the lack of ACID makes it impossible to serialize and replicate updates deterministically.
Ever since 7.1, I'm a near-fanatic PostgreSQL-user, if not to say fanboi.
I love, love, love PostgreSQL for its performance, its features, its awesome community and just for the fact that it's getting better and better every year.
Congratulations to the PostgreSQL team.
I've said it before and I'll probably say it many more times: You guys rule.
Sorry. This had to be said :-)