At 07:43 25/09/2011, Reuven M. Lerner wrote:
Hi, everyone. Daniel Verite
mailto:dan...@manitou-mail.orgdan...@manitou-mail.org wrote:
It would thus appear that there's a slight edge
for dumping bytea, but nothing
super-amazing. Deleting, however, is still
much faster with bytea than
Hi, everyone. Daniel Verite dan...@manitou-mail.org
wrote:
How much bytea are you dumping for it to take only 0.066s?
The fact that it takes about the same time than dumping the "empty content"
looks very suspicious.
On my desktop machine, if I create a
Reuven M. Lerner wrote:
When a record in the main table is deleted, there is a rule (yes a rule --
not a trigger) in the
referencing table that performs a lo_unlink on the associated object.
I just want to check that my intuition is correct: Wouldn't it be way faster
and more efficient
Hi, everyone. Albe wrote:
Could you try with a trigger instead of a rule and see if the
performance is better? Yours,
Laurenz Albe
Great idea. I did that, and here are the results for 10,000
records:
| | Delete |
Reuven M. Lerner wrote:
1.1 1,000 records
==
DeleteDump
---+-+
Empty content 0.172s0.057s
bytea 0.488s0.066s
large object30.833s 9.275s
How much bytea are you dumping
Hi, everyone. I'm working with someone who has a database
application currently running under PostgreSQL 8.3. Among other
things, there is a main table that is referenced by a number of
other tables via a foreign key. One of those tables has a field
of type
Hi again, everyone. I'm replying to my own posting, to add some
information: I decided to do some of my own benchmarking. And if my
benchmarks are at all accurate, then I'm left wondering why people
use large objects at all, given their clunky API and their extremely
