[pgpointcloud] Toast(ed)

[Paul Ramsey]

Actually not easily, though a function could be written to do that. Just checking the PC_MemSize and comparing w/ 8K is a rough/ready way.

P

-- 
Paul Ramsey
http://cleverelephant.ca
http://postgis.net

On February 7, 2014 at 12:10:27 AM, Roger Bedell (rbedell at coordinatesolutions.com) wrote:

Good morning Paul,
Is there any way to tell if your data got Toasted? (who made up that acronym?). I used 400 point patches, and everything seems fine so far, but I only have billions, not trillions of points.
Roger


On Thu, Feb 6, 2014 at 7:21 PM, Paul Ramsey <pramsey at cleverelephant.ca> wrote:
> Since we store patches with practically no extra information next to them in the tuple, just reading the PC_MemSize (I think that’s the function) of the patch gives an idea of storage you can use to bump up your patch size to close to the maximum (8kb). 

Is there a way we could communicate from the db to drivers.pgpointcloud.writer an automatic toast'able patch size based on this info? 
Yes, no, maybe so. You already have the core info (before looking at data) in your pipeline. The schema gives you an implicit point size. All you don’t know is the compression ratio, and until you start compressing actual data you don’t know that.

> You can also recompile your database with a higher page size if you’re feeling hacky. 

Is there any possibility that pg could be improved to allow tables of varying page sizes? I suppose increasing the page size generally across an entire database could cause some unforeseen pathologies. 
Next you’ll want to start f’ing with the gravitational constant. If you want to make the experiment, the page size is just a #define, and it might actually be a configure option even. It does cover the whole database though, variable page sizes aren’t on.

P.




_______________________________________________
pgpointcloud mailing list
pgpointcloud at lists.osgeo.org
http://lists.osgeo.org/cgi-bin/mailman/listinfo/pgpointcloud



--
Roger Bedell
Coordinate Solutions Inc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/pgpointcloud/attachments/20140207/2b32e95b/attachment.html>