Re: Load Distributed Checkpoints, take 3

On Mon, 25 Jun 2007, Tom Lane wrote:

> I'm not sure why you hold such strong allegiance to the status quo.  We
> know that the status quo isn't working very well.

Don't get me wrong here; I am a big fan of this patch, think it's an
important step forward, and it's exactly the fact that I'm so shell
shocked from abuse by the status quo that I'm still mixed up in this mess
(I really should be ignoring the lot of you and writing new code instead).

LDC certainly makes things better in almost every case.  My "allegiance"
comes from having seen a class of transactions where LDC made things worse
on a fast/overloaded system, in that it made some types of service
guarantees harder to meet, and I just don't know who else might run into
problems in that area.  I'm worried that if it's not adjustable, you're
introducing a risk that you'll take a step backward for some of this
code's users, and that will be hard to undo given the way releases are
structured here.

I spent some time trading stocks for a living.  There are sometimes
situations you can get into there where there is a tiny chance that
something very bad can happen with a trade, and many people get wiped out
by such things.  If it's possible in that situation to remove that risk
with something inexpensive, you do it, even though the net expected value
of the change might be slightly negative.  This seems like such a
situation to me.  If it's possible to take away the risk of other people
running into an unforseen problem with the LDC patch just by keeping a
knob that's already there, unless that's an expensive operation my opinion
is that you should pick a good default but not remove it yet.

> And if you think that the current code had enormous amounts of testing
> before it went in, I've got to disillusion you :-(

It's having been on the painful receiving end of that fact that makes me
so paranoid now :)

--
* Greg Smith gsmith@gregsmith.com http://www.gregsmith.com Baltimore, MD