PostgreSQL Keeps Evolving: Better Performance, Irix Dies

[phoronix]

Phoronix: PostgreSQL Keeps Evolving: Better Performance, Irix Dies

For those not closely watching the latest upstream developments of the PostgreSQL database server, a number of changes have been made recently to advance this major open-source project...

http://www.phoronix.com/vr.php?view=MTUxNzE

[Veerappan]

I had honestly thought that IRIX had been killed off by now, but it's still sad to hear. I spent a few years in College as the admin of an SGI lab that was used by an Organic Chem professor for visualization purposes in his classes. The CPUs in those machines (O2's) were slow, but the memory/graphics capabilities were outstanding when those machines were new.

[darkixion]

And those are kind of the tip of the iceberg for anyone who hasn't been following PostgreSQL recently. Just in PostgreSQL 9.1 - 9.3 there's been:

• writeable common table expressions
• foreign data wrappers (with writeable support, effectively mount any type of data source as a table)
• range types (e.g. date ranges, integer ranges, IP address ranges etc.)
• JSON support
• LATERAL support
• POSIX shared memory (no more messing around with SHMMAX)
• materialized views
• column-level collations
• new lightweight key locks for foreign keys
• cascading streaming replication
• index-only scans
• unlogged tables
• synchronous replication
• page checksums
• event triggers

... along with many more.

And coming soon:

• parallel query
• minmax indexes
• logical replication
• concurrent reindexing
• faster-than-MongoDB JSON indexing
• ultra-fast tuple freezing
• BEFORE/AFTER support on RETURNING clause
• time-delayed standbys

Incidentally, I caused a performance patch to be reverted as I found it was subject to a bug with extents in the EXT4 filesystem. Unfortunately we couldn't get feedback from Theodore Ts'o on that.

Thom