Accessing DBMS remotely: MySQL? FireBird?

[Fred the man]

Hi,

Some of our customers have remote offices. I was wondering if it'd be
safe to have a DBMS running at their central office, and have our
client application running on hosts in the branches connect to it
through a VPN via the Net?

What happens if the connection goes south while a branch office was
making changes? Does the DBMS just rollbacks changes automatically
after a time-out?

Should we set up some kind of replication instead?

Also, are there compeling reasons to go for Firebird instead of MySQL?
I don't know enough about the capabilities of each DBMS today to make
an educated choice.

Thank you for any tip
Fred.

[bobb]

couple of notes....
Sounds like the app's not browser based.... ( that may take care of your
problem...)
how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).
Don't forget... if data not sensitive, you may not need a vpn. (SLOOOWWW).
Replication models should be based on the "just on time-liness" need of the
data.
do they need it REALtime, NEARtime, or tomorrow? That's the question that
should give you your
replication answer (IMO)
"Fred the man" <fr**@acme.com> wrote in message
news:a1********************************@4ax.com...

Hi,

Some of our customers have remote offices. I was wondering if it'd be
safe to have a DBMS running at their central office, and have our
client application running on hosts in the branches connect to it
through a VPN via the Net?

What happens if the connection goes south while a branch office was
making changes? Does the DBMS just rollbacks changes automatically
after a time-out?

Should we set up some kind of replication instead?

Also, are there compeling reasons to go for Firebird instead of MySQL?
I don't know enough about the capabilities of each DBMS today to make
an educated choice.

Thank you for any tip
Fred.

[bobb]

couple of notes....
Sounds like the app's not browser based.... ( that may take care of your
problem...)
how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).
Don't forget... if data not sensitive, you may not need a vpn. (SLOOOWWW).
Replication models should be based on the "just on time-liness" need of the
data.
do they need it REALtime, NEARtime, or tomorrow? That's the question that
should give you your
replication answer (IMO)
"Fred the man" <fr**@acme.com> wrote in message
news:a1********************************@4ax.com...

Hi,

Some of our customers have remote offices. I was wondering if it'd be
safe to have a DBMS running at their central office, and have our
client application running on hosts in the branches connect to it
through a VPN via the Net?

What happens if the connection goes south while a branch office was
making changes? Does the DBMS just rollbacks changes automatically
after a time-out?

Should we set up some kind of replication instead?

Also, are there compeling reasons to go for Firebird instead of MySQL?
I don't know enough about the capabilities of each DBMS today to make
an educated choice.

Thank you for any tip
Fred.

[OneSolution]

Also, if you use InnoDB tables, then you won't have a problem if the link
fails provided you use transactions for the SQL. So you turn off auto
commit and then perform the transaction. If there is a network failure,
then the commit never happens and you can simply re-issue the command when
the network link comes back up.
"bobb" <ra*****@hotmail.com> wrote in message
news:3T******************@nwrddc02.gnilink.net...

couple of notes....
Sounds like the app's not browser based.... ( that may take care of your
problem...)
how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).
Don't forget... if data not sensitive, you may not need a vpn. (SLOOOWWW).
Replication models should be based on the "just on time-liness" need of the data.
do they need it REALtime, NEARtime, or tomorrow? That's the question that
should give you your
replication answer (IMO)
"Fred the man" <fr**@acme.com> wrote in message
news:a1********************************@4ax.com...

Hi,

Some of our customers have remote offices. I was wondering if it'd be
safe to have a DBMS running at their central office, and have our
client application running on hosts in the branches connect to it
through a VPN via the Net?

What happens if the connection goes south while a branch office was
making changes? Does the DBMS just rollbacks changes automatically
after a time-out?

Should we set up some kind of replication instead?

Also, are there compeling reasons to go for Firebird instead of MySQL?
I don't know enough about the capabilities of each DBMS today to make
an educated choice.

Thank you for any tip
Fred.

[OneSolution]

Also, if you use InnoDB tables, then you won't have a problem if the link
fails provided you use transactions for the SQL. So you turn off auto
commit and then perform the transaction. If there is a network failure,
then the commit never happens and you can simply re-issue the command when
the network link comes back up.
"bobb" <ra*****@hotmail.com> wrote in message
news:3T******************@nwrddc02.gnilink.net...

couple of notes....
Sounds like the app's not browser based.... ( that may take care of your
problem...)
how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).
Don't forget... if data not sensitive, you may not need a vpn. (SLOOOWWW).
Replication models should be based on the "just on time-liness" need of the data.
do they need it REALtime, NEARtime, or tomorrow? That's the question that
should give you your
replication answer (IMO)
"Fred the man" <fr**@acme.com> wrote in message
news:a1********************************@4ax.com...

Hi,

Some of our customers have remote offices. I was wondering if it'd be
safe to have a DBMS running at their central office, and have our
client application running on hosts in the branches connect to it
through a VPN via the Net?

What happens if the connection goes south while a branch office was
making changes? Does the DBMS just rollbacks changes automatically
after a time-out?

Should we set up some kind of replication instead?

Also, are there compeling reasons to go for Firebird instead of MySQL?
I don't know enough about the capabilities of each DBMS today to make
an educated choice.

Thank you for any tip
Fred.

[Fred the man]

On Sun, 02 May 2004 14:48:31 GMT, "bobb" <ra*****@hotmail.com> wrote:

Sounds like the app's not browser based.... ( that may take care of your
problem...)
Indeed. It's written in VB. Until now, customers only used our apps in
a LAN, but some of them now have branches, hence the problem of
accessing data through ADSL with a VPN between the two locations.
how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).
512Kbps at the branch, 1024Kbps at the main office. FWIW, Terminal
Services works very well, which is what I recommended they set up
until we sort things about with a DBMS.
do they need it REALtime, NEARtime, or tomorrow? That's the question that
should give you your >replication answer (IMO)

They make changes to records all day long, so real/near time is a
requirement.

Thank you
Fred.

[Fred the man]

On Sun, 02 May 2004 14:48:31 GMT, "bobb" <ra*****@hotmail.com> wrote:

Sounds like the app's not browser based.... ( that may take care of your
problem...)
Indeed. It's written in VB. Until now, customers only used our apps in
a LAN, but some of them now have branches, hence the problem of
accessing data through ADSL with a VPN between the two locations.
how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).
512Kbps at the branch, 1024Kbps at the main office. FWIW, Terminal
Services works very well, which is what I recommended they set up
until we sort things about with a DBMS.
do they need it REALtime, NEARtime, or tomorrow? That's the question that
should give you your >replication answer (IMO)

They make changes to records all day long, so real/near time is a
requirement.

Thank you
Fred.

[Fred the man]

On Mon, 03 May 2004 08:30:39 GMT, "OneSolution"
<on*********@sbcglobal.net> wrote:

Also, if you use InnoDB tables, then you won't have a problem if the link
fails provided you use transactions for the SQL. So you turn off auto
commit and then perform the transaction. If there is a network failure,
then the commit never happens and you can simply re-issue the command when
the network link comes back up.

Cool :-) The documentation didn't make that clear, since they
mentionned isueing the "rollback" command, so I was wondering if it
could be done automagically when a connection is severed halfway
through a transaction.

Thank you
Fred.

[Fred the man]

On Mon, 03 May 2004 08:30:39 GMT, "OneSolution"
<on*********@sbcglobal.net> wrote:

Also, if you use InnoDB tables, then you won't have a problem if the link
fails provided you use transactions for the SQL. So you turn off auto
commit and then perform the transaction. If there is a network failure,
then the commit never happens and you can simply re-issue the command when
the network link comes back up.

Cool :-) The documentation didn't make that clear, since they
mentionned isueing the "rollback" command, so I was wondering if it
could be done automagically when a connection is severed halfway
through a transaction.

Thank you
Fred.

[bobb]

"Fred the man" <fr**@acme.com> wrote in message
news:eo********************************@4ax.com...

On Sun, 02 May 2004 14:48:31 GMT, "bobb" <ra*****@hotmail.com> wrote:

Sounds like the app's not browser based.... ( that may take care of your
problem...)
Indeed. It's written in VB. Until now, customers only used our apps in
a LAN, but some of them now have branches, hence the problem of
accessing data through ADSL with a VPN between the two locations.

how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).

512Kbps at the branch, 1024Kbps at the main office. FWIW, Terminal
Services works very well, which is what I recommended they set up
until we sort things about with a DBMS.

Yea, I'd think that'd be the way to go. I'm not a big fan of terminal
services (as much as
I was anyway). Without it though, you'd either have to do local app and
then connect (odbc, ado, whatever)
remotely OR worser, remote app + remote database. At least with terminal
services
you take the time-out out of the picture. imo.
good luck.
bobb

do they need it REALtime, NEARtime, or tomorrow? That's the question thatshould give you your >replication answer (IMO)

They make changes to records all day long, so real/near time is a
requirement.

Thank you
Fred.

[bobb]

"Fred the man" <fr**@acme.com> wrote in message
news:eo********************************@4ax.com...

On Sun, 02 May 2004 14:48:31 GMT, "bobb" <ra*****@hotmail.com> wrote:

Sounds like the app's not browser based.... ( that may take care of your
problem...)
Indeed. It's written in VB. Until now, customers only used our apps in
a LAN, but some of them now have branches, hence the problem of
accessing data through ADSL with a VPN between the two locations.

how fast is the link between the offices? I'm sure it's over the internet
right? (no private network).

512Kbps at the branch, 1024Kbps at the main office. FWIW, Terminal
Services works very well, which is what I recommended they set up
until we sort things about with a DBMS.

Yea, I'd think that'd be the way to go. I'm not a big fan of terminal
services (as much as
I was anyway). Without it though, you'd either have to do local app and
then connect (odbc, ado, whatever)
remotely OR worser, remote app + remote database. At least with terminal
services
you take the time-out out of the picture. imo.
good luck.
bobb

do they need it REALtime, NEARtime, or tomorrow? That's the question thatshould give you your >replication answer (IMO)

They make changes to records all day long, so real/near time is a
requirement.

Thank you
Fred.