*.osm or *.pbf ?
by Pascal Neis - Published: October 23rd, 2010
Since September 5 th Osmosis supports the new OSM binary fileformat. It sounds interesting, but where are the pros of this format?
I played a little bit with the OSM file of entire Europe. The europe (*.osm) file has an uncompressed format size of about 72.9 GB (compressed it is about 5.2 GB). The new OSM binary (*.pbf) file on the other hand has a size of 3.7 GB (compress=deflate) or 7.6 GB (compress=none).
With the help of Osmosis, it’s quite simple to update an OSM file daily via the “diff” files. You can find a good “how to” in the OSM wiki (here).
For the past 5 days, I collected the processing times that Osmosis (version 0.37) takes to update the europe *.osm file. The osmosis job contains the download of the change (*.osc) file and the cutting (bounding-polygon parameter) of Europe. Altogether the job runs at average in 56min. With the OSM *.pbf file the same task is completed in 14min. I think this is a big difference. So if you need an OSM file on your system, give the new binary OSM files a try! Really nice work Scott 🙂
Some system information: (i7-920) Quadcore with 4x 2667MHz, 1500GB HDD and 12GB RAM. thx @ dennis
[…] This post was mentioned on Twitter by Jonas K. and Pascal Neis, Tisham Dhar. Tisham Dhar said: RT @pascal_n: New blog post: *.osm or *.pbf ? http://neis-one.org/2010/10/23/osm-or-pbf/ #OSM […]
Hi Pascal, nice benchmark, thank you 🙂
But as some other people already noted, the new binary format is a little bit more magic than our easy to read XML. Technicaly there are a lot of pros but we should keep in mind that there are beginners, too. And that ones need an easy way to join OSM data processing and development using OSM XML in future 🙂
Whenever openstreetmap introduces a new file format, a lot of applications need to change to understand a new file format.
openstreetmap should finalize the new file format .pbf definition before introduction it to public, to avoid a lot of changes need to be done for downstream applications.
[…] интернете я наткнулся на заметку *.osm or *.pbf ?, в которой сравнивались размеры файлов в этих […]