Re: star join optimization

On 11/14/2011 10:44 AM, Rudyar wrote:
> Ok, I'm working in that project. I will send you my results
> and comparision with SQL server HHJ optimization in one or two months.

Please be careful not to share here details of how features like this
are built in any commercial databases you evaluate. Some of those
implementations use patented design approaches that should be avoided in
an open source project. Oracle, Microsoft, and DB2 are all aggressive
about patenting the innovative parts of their database server code.

In addition to not wanting to accidentally incorporate such a design,
it's better for the PostgreSQL project to not be aware of what patents
in this area exist too. We don't even want a survey of patents in this
area published here because there are increased penalties for willful
patent infringement. See http://en.wikipedia.org/wiki/Treble_damages
for example.

What this project likes best are innovative approaches from recent
academic research that haven't been incorporated in any commercial
products yet. A good example is how the Serializable Snapshot Isolation
technique developed by Cahill and others was added to PostgreSQL 9.1:
http://wiki.postgresql.org/wiki/Serializable There was less concern
over accidentally duplicating a patented approach because that technique
wasn't in any of the commercial databases yet.

--
Greg Smith 2ndQuadrant US greg(at)2ndQuadrant(dot)com Baltimore, MD
PostgreSQL Training, Services, and 24x7 Support www.2ndQuadrant.us