[OSM-dev] Osmosis multi-task questions

[Lambertus]

Martijn van Oosterhout wrote:
> On Nov 27, 2007 10:14 AM, Lambertus <osm at na1400.info> wrote:
>> I replied to an offlist mail from Frederik about this, but I would like
>> to add some thoughts here too, now that I see this topic on the ML.
> 
>> However, keeping that Europe file updated with daily diffs seems like a
>> clean method to prevent plowing through a lot of useless data every week.
> 
> Keeping a whole planet file up-to-date is fairly quick, hypercube has
> been generating daily planet dumps for a while now and it takes about
> an hour. Next to that it's generating daily NL and scandinavia dumps.
> I havn't really thought about optimisation given hypercube can extract
> a bbox from the planet dump in about 20 minutes, so it'd need to do a
> lot more countries before it became an issue. Ay that point I'll try
> the --tee option to see how it compares...
> 
> The relevent scripts are here: http://hypercube.telascience.org/planet/
> 

My tests reveal:
Extracting bounding boxes one-by-one from the planet file uses: tN. 
Where 't' is the time to extract one bbox, 'N' is the amount of bboxes.

Extracting multiple bboxes from the planet file using 'tee' takes less 
then: 2t. Where 't' is the time to extract one bbox.

So when extracting more then 1 bbox I suggest using the 'tee' option.