Re: How are locks managed in PG?

On Sun, Dec 21, 2008 at 8:48 PM, Jonah H. Harris <jonah.harris@gmail.com> wrote:
> On Sun, Dec 21, 2008 at 9:42 PM, David Fetter <david@fetter.org> wrote:
>> On Sun, Dec 21, 2008 at 08:46:15PM -0500, Jonah H. Harris wrote:
>>> On Fri, Dec 19, 2008 at 7:49 AM, Alvaro Herrera
>>> <alvherre@commandprompt.com> wrote:
>>> >> Oracle on the other hand stores the lock information directly in
>>> >> the data block that is locked, thus the number of locks does not
>>> >> affect system performance (in terms of managing them).
>>> >>
>>> >> I couldn't find any description on which strategy PG applies.
>>> >
>>> > None of the above.  We're smarter than everyone else.
>>>
>>> Which is why Oracle's locks are more scalable than PG's?
>>
>> You've been talking about your super-secret test which you allege,
>> quite implausibly, I might add, to have Oracle (8i, even!) blowing
>> PostgreSQL's doors off for weeks now.
>>
>> Put up, or shut up.
>
> Same to the standard PG B.S. responses such as, "None of the above.
> We're smarter than everyone else."  When's the last time Alvaro used
> or tuned Oracle?  Does he have a clue about how Oracle locks scale?
> Stop complaining.

The difference is HE put forth an opinion about the pg developers
being smarter, but you put forth what seems like a statement of fact
with no evidence to back it up.  One is quite subjective and open for
debate on both sides, and often to good effect.  The other is a
statement of fact regarding scalability in apparently all usage
circumstances, since it wasn't in any way clarified if you were
talking about a narrow usage case or all of the possible and / or
probably ones.

Having dealt with cust service for a few commercial dbs, I can safely
say I get way better service from way smarter people when I have a
problem.  And I don't have a lot of problems.