I went geocaching this morning, because I have a Travel Bug to drop in Canada. But I only found a pile of rubble. Ran into Bruce just getting out of his cab from the airport on the way back.
I presented my tutorial on porting Oracle applications to PostgreSQL in the afternoon. This went quite well, and I received encouraging feedback afterwards. The slides of the presentation should appear on the PGCon web site at the linked URL. For those who were confused by the somewhat cynical tone of the presentation: I have been traumatized a bit by the issue. By and large many applications are quite easy and straightforward to port. I certainly do encourage these efforts.
One thing that came up after the presentation that I have not considered in great depth is the issue of performance of the ported result. In the discussion, a few possibilities were mentioned:
- The Oracle application is so carefully tuned with optimizer hints, it will never perform on PostgreSQL. We probably can't/won't port it.
- Half the time of a porting project will be required to tune the PostgreSQL port, because Oracle optimizes bad queries much better.
- Some things perform better in PostgreSQL, some worse. It probably averages out.
- With the money you save with PostgreSQL, you can afford better hardware.
More insight on this issue would be welcome.
I think I'll go on a second geocaching attempt now and hit the Royal Oak pub later with the rest of the group.
Note for those coming from across the pond: The Champions League final is televised on TSN and RDS beginning at 14:00 tomorrow. You get both of these channels in the university residence. But there's the developer meeting ...