On 06/27/2013 11:13 PM, Jeff Janes wrote:
> Wouldn't any IO system being used on a high-end system be fairly good
> about making this work through interleaved read-ahead algorithms?
To some extent, certainly. It cannot possibly get better than a fully
sequential load, though.
> That sounds like it would be much more susceptible to lock contention,
> and harder to get bug-free, than dividing into bigger chunks, like whole
> 1 gig segments.
Maybe, yes. Splitting a known amount of work into equal pieces sounds
like a pretty easy parallelization strategy. In case you don't know the
total amount of work or the size of each piece in advance, it gets a bit
harder. Choosing chunks that turn out to be too big certainly hurts.
Regards
Markus Wanner