Subject | Re: [Firebird-Architect] None local database file |
---|---|
Author | Jim Starkey |
Post date | 2003-07-30T16:07:10Z |
Klement Guenther - Munich-MR wrote:
need to be considered.
The historical reason for the restriction was that it was impossible to
prevent or detect two servers (or classic clients) from going at the
database from different nodes, which is an absolutely certain way to
trash a database beyond recovery. On networks that could reliably
detect cross-node access (Apollo domain), we let anybody attach any
database, and subsequent attachments would automatically go remote to
wherever the database was active. We started, but didn't complete, a
page/lock server for other networks.
There are those who argue that users should be given enough rope to hang
themselves, but those people generally don't have to support the corpses.
>It should be possible to have the database file on any location the server have access to.That's a very good principle, but sometimes operational considerations
>To restrict database files to local drives makes no sense as Gigabit netzwork is todays standard.
>It should always be up to the user where to put the database file and not forced be implementation.
>
>
need to be considered.
The historical reason for the restriction was that it was impossible to
prevent or detect two servers (or classic clients) from going at the
database from different nodes, which is an absolutely certain way to
trash a database beyond recovery. On networks that could reliably
detect cross-node access (Apollo domain), we let anybody attach any
database, and subsequent attachments would automatically go remote to
wherever the database was active. We started, but didn't complete, a
page/lock server for other networks.
There are those who argue that users should be given enough rope to hang
themselves, but those people generally don't have to support the corpses.