| Subject | Re: [ib-support] Firebird and Network Attached Storage? |
|---|---|
| Author | Artur Anjos |
| Post date | 2002-01-15T21:32:15Z |
David:
Replication should be the alternative, but it depends on the application.
I have implemented a solution that it's not very expensive, and the users
looses maximum 'just' one hour of work if something went wrong with the main
server (30/40 users online). I have two servers: on the main server, I do a
backup each hour, and ftp it to the 'backup' server. I have an agent on the
backup server that receives the backup and restore it. If the main server
fails for any reason, one of the users went there, shutdown the main server
and change the backup server IP to the one of the main server. In this
situation, the users loose maximum 1 hour of work (if the main server fails
after 10m of working, they will just loose 10m). This is a small 350Mb
database, so everything it's fast.( the back up, zip and ftp takes about 3
minutes). I have 2 nics in each server: one connected to main switch,
another with a crossover cable between the 2 machines, just to speed up ftp
transfer.
It's not replication, but it's a solution. A cheap solution, with a defined
'time lost' production.
I keep a small agent on the main server looking at some database changes and
creating a log for some changes that could not be verified easily if
something fails in a hour (such as the serial numbers of the units that are
leaving the wharehouse): this log runs every 10 minutes, and is also ftp to
the backup server.
Another advantages: You really have backups of the database (The agent on
the backup server mail to everyone if it didn't received a backup, and mail
twice a day to someone saying that it did).
What I'm trying to saying here it's that we can find alternative solutions.
Artur
Replication should be the alternative, but it depends on the application.
I have implemented a solution that it's not very expensive, and the users
looses maximum 'just' one hour of work if something went wrong with the main
server (30/40 users online). I have two servers: on the main server, I do a
backup each hour, and ftp it to the 'backup' server. I have an agent on the
backup server that receives the backup and restore it. If the main server
fails for any reason, one of the users went there, shutdown the main server
and change the backup server IP to the one of the main server. In this
situation, the users loose maximum 1 hour of work (if the main server fails
after 10m of working, they will just loose 10m). This is a small 350Mb
database, so everything it's fast.( the back up, zip and ftp takes about 3
minutes). I have 2 nics in each server: one connected to main switch,
another with a crossover cable between the 2 machines, just to speed up ftp
transfer.
It's not replication, but it's a solution. A cheap solution, with a defined
'time lost' production.
I keep a small agent on the main server looking at some database changes and
creating a log for some changes that could not be verified easily if
something fails in a hour (such as the serial numbers of the units that are
leaving the wharehouse): this log runs every 10 minutes, and is also ftp to
the backup server.
Another advantages: You really have backups of the database (The agent on
the backup server mail to everyone if it didn't received a backup, and mail
twice a day to someone saying that it did).
What I'm trying to saying here it's that we can find alternative solutions.
Artur
----- Original Message -----
From: "David Montgomery" <montgomery_list@...>
To: <ib-support@yahoogroups.com>
Sent: Tuesday, January 15, 2002 8:47 PM
Subject: RE: [ib-support] Firebird and Network Attached Storage?
> Fred,
>
> If that is the case, then how do other IB/FB installations provide some
type
> of fail-over redundancy? Replication seems to be the only other
> alternative, and with that scenario there would seem to need to be two
> roughly comparable IB/FB servers sitting side-by-side, which seems like a
> very expensive option...
>