PostgreSQL 12 Beta Released With Performance Improvements

[phoronix]

Phoronix: PostgreSQL 12 Beta Released With Performance Improvements

Out today is the first beta of the upcoming PostgreSQL 12.0 database server...

http://www.phoronix.com/scan.php?page=news_item&px=PostgreSQL-12-Beta-Released

[hajj_3]

I hope that you can benchmark it Michael compared to v11.x

[uid313]

Seems like a really great database. Unfortunately it puts lots of files on the file system like "createuser", etc, very confusing.
Instead of prefixing all binaries with pg_ or using a pg wrapper script like the "git" and "dotnet" commands does.

[polarathene]

Seems like a really great database. Unfortunately it puts lots of files on the file system like "createuser", etc, very confusing.
Instead of prefixing all binaries with pg_ or using a pg wrapper script like the "git" and "dotnet" commands does.

Use a container, problem solved!

[jrch2k8]

Seems like a really great database. Unfortunately it puts lots of files on the file system like "createuser", etc, very confusing.
Instead of prefixing all binaries with pg_ or using a pg wrapper script like the "git" and "dotnet" commands does.

Yea it does but remember PostgreSQL was designed originally to see eye to eye with Oracle(IBM DB2, etc.) and many of those annoyances exists to make Oracle(IBM DB2, etc.) migration less painful for DBAs.

You could give the same reasoning for some of their choices with PLSQL and other toolings

[jrch2k8]

I hope that you can benchmark it Michael compared to v11.x

Just saying here, if you need to see benchmarks for PostgreSQL :

1.) You don't need it, use MySQL instead and save yourself a ton of trouble.
2.) Any result you get is absolutely useless and probably impossible to reproduce in 99% of the cases that would be useful for anybody that actually need PostgreSQL.

I don't mean any of this as an insult or anything, i literally mean PostgreSQL is a 10000 heads hydra(like Oracle/DB2/etc.) and literally everything will wildly affect any result, anything from a really simple parameter at postgresql.conf all the way to 1 field type on one of your 10000 tables will make a difference between processing 100 IOPS or 1000000 IOPS this is without actually taking hardware into account which is another massive monster on its own.

In layman terms:

MySQL == JavaScript high level framework
PostgreSQL == low level kernel like C with inline assembly

[andrei_me]

Just saying here, if you need to see benchmarks for PostgreSQL :

1.) You don't need it, use MySQL instead and save yourself a ton of trouble.
2.) Any result you get is absolutely useless and probably impossible to reproduce in 99% of the cases that would be useful for anybody that actually need PostgreSQL.

I don't mean any of this as an insult or anything, i literally mean PostgreSQL is a 10000 heads hydra(like Oracle/DB2/etc.) and literally everything will wildly affect any result, anything from a really simple parameter at postgresql.conf all the way to 1 field type on one of your 10000 tables will make a difference between processing 100 IOPS or 1000000 IOPS this is without actually taking hardware into account which is another massive monster on its own.

In layman terms:

MySQL == JavaScript high level framework
PostgreSQL == low level kernel like C with inline assembly

Ohhh, what an insult, most common perf pitfalls can be check through analyze.

When an index is not used, you can check things like default_statistics_target or per column statistics, you check type mismatches or collate used on the column.

I, for one, dislike MySQL

[jrch2k8]

Ohhh, what an insult, most common perf pitfalls can be check through analyze.

When an index is not used, you can check things like default_statistics_target or per column statistics, you check type mismatches or collate used on the column.

I, for one, dislike MySQL

i mostly agree but once a while ago i knew a badass real PostgreSQL DBA expert and understood that i knew nothing but yeah analyze is the bare basic to fix oblivious performance booboos

[Jonjolt]

Just saying here, if you need to see benchmarks for PostgreSQL :

1.) You don't need it, use MySQL instead and save yourself a ton of trouble.
2.) Any result you get is absolutely useless and probably impossible to reproduce in 99% of the cases that would be useful for anybody that actually need PostgreSQL.

I don't mean any of this as an insult or anything, i literally mean PostgreSQL is a 10000 heads hydra(like Oracle/DB2/etc.) and literally everything will wildly affect any result, anything from a really simple parameter at postgresql.conf all the way to 1 field type on one of your 10000 tables will make a difference between processing 100 IOPS or 1000000 IOPS this is without actually taking hardware into account which is another massive monster on its own.

In layman terms:

MySQL == JavaScript high level framework
PostgreSQL == low level kernel like C with inline assembly

Unless you can't use PostgreSQL because of availability MySQL and it's derivatives are a dumpster fire, I am working on an ETL program transferring stuff from MariaDB that uses MyISAM (Legacy Stuff) to Elasticsearch importing the dataset to PostgreSQL (Unmodified default config) cuts the indexing time down to less than a third. Using an Intel 750 PCIe HD.

[dkasak]

Just saying here,
,
... <blah blah> ...

In layman terms:

MySQL == JavaScript high level framework
PostgreSQL == low level kernel like C with inline assembly

Nah, sorry. I've been DBA ( and backend dev, and ETL dev ) using both MySQL and Postgres over the years, and while there are differences, it's nothing like you say. Maybe you got in over your head on a Postgres project?