Alexandre Benson Smith wrote:

> ...
> I think a bit of experimentation will be necessary.
>
> What I could say is:
> 1.) Blob Pages are not shared between blob's, so the tendency is to have
> more unused/wasted space with big page sizes (ok, HD space is not
> important nowadays)
> 2.) If a Blob is small enough to fit in the data page it will be, so
> less records will be stored on the same page which increases the number
> of pages to be read (So small page size should avoid this)
>
> But I would go for 16kb page, and have a test for performance and size
> on 8kb and probably 4kb too and see if it's significant.
>

Since very few images are less than 16KB (they are mostly 50KB and
more) most blobs would span several pages and I reasoned that the
fewer pages the better. If an image is 90KB, at least 6 pages (@
16KB) must be read, while for 8KB pages at least 12 pages will be
read. Will the total time for reading the pages only depend on the
blob size, and be independent of the number of pages?



BTW:
Testing backup and restore of the database with 140.000 images -
7.5GB - which is about 1/40 of the total shows that backup takes
about 7-8 minutes (and so did restore). This means that backup (or
restore) of the full database will take about 5-6 hours. No problem
at all. (Don't worry - be happy!).
The restored database was just 1MB smaller than the original (7.5GB total).

--
Aage J.