> - hard transactions: cutting the power to a PostgreSQL server
> may cause data loss and/or data corruption. Some databases use
> sophisticated techniques to ensure serialization of operation
> through journaling, redoing some of the transactions at
> bootup time if required.
>
> - ability to have databases bigger than the host's maximum file size
Neither of these are problems in PostgreSQL. Tom Lane already answered the
transaction one, but I'll answer the size one. PostgreSQL stores each table
in a separate file, and will store a table in multiple files if it becomes
larger than about 1 gig. I've had (for testing, not production) databases
with 30 GB of data on Linux where the file limit was 2 GB.
As for your other two concerns, I'd like to see them fixed as well. I
handle those issues currently at the application level, but I'm considering
writing a small change to the core to provide that functionality (as soon as
I finish my other add on projects).
--
Adam Ruth
InterCation, Inc.
www.intercation.com
"Marc SCHAEFER" <schaefer@alphanet.ch> wrote in message
news:Pine.LNX.3.96.1001010155021.1563A-100000@defian.alphanet.ch...
> Hi,
>
> I have been using PostgreSQL a bit and I have been surprised by the good
> quality, the features, and now the Addison-Wesley book which is excellent.
> However, I have some questions about the implementation of a few
> additional features.
>
> I suppose that PostgreSQL hasn't any ability to do the following yet:
>
> - log all transactions to a special log file, that can be used for
> backup purposes: ie you dump the database every day, but you keep
> the transaction log on a separate disk. Should the database disk
> crash, you won't have any data loss if you restore the backup and
> replay the transaction log.
>
> - hard transactions: cutting the power to a PostgreSQL server
> may cause data loss and/or data corruption. Some databases use
> sophisticated techniques to ensure serialization of operation
> through journaling, redoing some of the transactions at
> bootup time if required.
>
> - the ability to synchronize two database servers, with only the
> changes being exchanged, live. Or the ability to have many
> servers in a load-balancing or data scattering pool.
>
> - ability to have databases bigger than the host's maximum file size
>
> if this is true, can someone tell me why it hasn't been implemented yet
> (there might be very good reasons), if it will be, and if the task seems
> complicated.
>
> thank you for your time.
>
>
>