Steve -


We pull TIFF's in batches, and the blob storage for these TIFF's is
not sequential. In other words, as is typical in any query, you are
pulling data that is not sequentially paged, so I guess it takes a
long time to find blobs like this when they are spread across several
pages.

Thanks for the info.

David Keith

--- In firebird-support@yahoogroups.com, Steve Wiser <steve@...> wrote:
>
> Hi David,
>
> We really don't do anything special... The binary blobs are created as
> sub_type 0 with a segment size of 4096. The database page size is
> 8192. The only thing special I guess is that we put the image db on 1
> server and the data db on a different server.
>
> We pull the images 1 at a time though based on the key so we don't
> really grab a whole bunch of TIFF or PDF files and send them over the
> network when we can help it. I would say that if you use a
generator or
> something for the PK and just pull blobs using it you should be
fine. I
> am not sure what other people do though.
>
> Does that help at all?
>
> -steve