Postgresql database for distributed transactions

Hi, I am designing a mobile POS application. I was hoping I can use the
Postgresql database as the central transactional database.

The application itself will run on a remote mobile PC. The retail
transaction data is small, it consists of the transaction timestamp,
location, amount, item price, item type, etc.

There are at least 600 of these types of mobile POS required. Each of
the mobile PCs DOES NOT have a stable and consistent network connection
throughout the day. For example, the network may be available for 5
minutes in the morning, and no connection at all in the afternoon.
However, there are at least 30 minutes of guaranteed stable connection
time at midnight every day.

Each of the mobile POS when facing the consumer needs to process at most
500 transactions a day. I was hoping to use the intermittent network
connection time to do system download to flag transactions, such as,
stolen credit card number, etc. And, to use the stable midnight network
connection time to do uploads of the mobile transactions to the central
Postgresql database. The upload would require at most 600 mobile
terminals times 500 transactions at most of 3,000,000 transactions a day
in 30 minutes time frame.

What specific Postgresql feature can I use to process the upload
transactions? should I attempt to write transaction logs or do bulk
inserts. What specific features in Postgresql would help to ensure the
data consistency in such a computing environment? Thanks in advance.

William Hu

Trimet.org