Re: can we add SKIP LOCKED to UPDATE?

On 12/11/15 02:07, Craig Ringer wrote:
>
>
> On 11 November 2015 at 16:02, Torsten Zühlsdorff 
> <mailinglists@toco-domains.de <mailto:mailinglists@toco-domains.de>> 
> wrote:
>
>     From my experience most databases are just tpo small. Their
>     operations finish before there can be a deadlock. Same for race
>     conditions - most developer don't know about them, because they
>     never stumbled upon them. I am matching regularly discussions if a
>     database is already to big when holding 10.000 records in the
>     whole cluster...
>
>
> Ha. Yes. So true.
>
> I see Stack Overflow posts where somebody explains that their query 
> takes ages on their Huge!!1! database. Then it turns out the query 
> takes 0.2 seconds on a 400MB table.
>
> Huge. Right.
>
> -- 
>  Craig Ringer http://www.2ndQuadrant.com/
>  PostgreSQL Development, 24x7 Support, Training & Services

I'll say its huge.

Don't you realize that 400MB is over 4 million of the old 100Kb floppy 
disks, and even with the new big 1.44MB 3.5 " disks, you'd need about 280!!!

Though, I suspect that the people who are saying that 400MB is huge, 
never actually used those floppies I mentioned above.

Now-a-days I would not regard 400GB as huge, even though when I started 
programming, MainFrames often had less than 1MB of core memory, and the 
big 12" tape reels could hold a max of 35MB of data.  Now I'm sitting 
sitting at a Linux box were even my SSD has hundreds of times the 
storage (let alone the capacity of my HD's) the entire New Zealand Post 
Office had in 1980!  How times change...

The reality, is that people tend to compare things in their direct 
experience, and don't have a feeling for the 'size' of the computers 
they use in terms of storage & processing power.  The above mainframe I 
mentioned (an ICL 4/72) had washing machine sized boxes each with a 60MB 
disk - everything was impressively sized, now you can fit a TB HD in a 
match box.


Cheers,
Gavin