Carl,
I don't have experience with that big databases, but I did both
solutions, and here are pros of both of them:
1. Files stored on the filesystem:
- Small database footprint
- Faster backup, export and import
2. Files stored in the database
- RDBMS takes care of transactions and ref. int.
- Slower backup, export and import but all done in one step
- Easier continuous archiving
I slightly prefer option no. 2, since transaction handling, rollback and
ref. integrity is not so easy to implement when you have two different
storage systems (FS and RDB).
As for indexes and tables it is not clear form your message whether you
need just a regular search (field LIKE 'something'), full text search of
metadata, or full text search of scanned documents (in case they are OCRed).
Regards,
Ognjen
On 1.6.2011 10:08, Carl von Clausewitz wrote:
Hello Everyone,
I got a new project, with 100 user in Europe. In this case, I need to
handle production and sales processes an its documentations in
PostgreSQL with PHP. The load of the sales process is negligible, but
every user produces 2 transaction in the production process, with 10-30
scanned documents (each are 400kb - 800kb), and 30-50 high resolution
pictures (each are 3-8 MB), and they wanted to upload it to 'somewhere'.
'Somewhere' could be the server files system, and a link in
the PostgreSQL database for the location of the files (with some
metadata), or it could be the PostgreSQL database.
My question is that: what is your opinion about to store the scanned
documentation and the pictures in the database? This is a huge amount of
data (between daily 188MB and 800MB data, average year is about 1 TB
data), but is must be searchable, and any document must be retrieved
within 1 hour. Every documentations must be stored for up to 5 years...
It means the database could be about 6-7 TB large after 5 years, and
then we can start to archive documents. Any other data size is negligible.
If you suggest, to store all of the data in PostgreSQL, what is your
recommendation about table, index structure, clustering, archiving?
Thank you in advance!
Regards,
Carl
--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
