Subject | Re: [firebird-support] pumping very large records (30 megs!) |
---|---|
Author | Artur Anjos |
Post date | 2003-08-16T15:27:26Z |
Spou,
With a slow connection, transfer 69gigs will be something that I will never
think about. :-)
Even with ZeBedde. If we are talking about images, sometimes the image
format is already compressed, so....
Here is a possible solution.
You will need to have some work, but this is an easy task with Delphi+IBO:
a) Pump the initial database as you are already doing, without the BLOB
data.
This transfer will be fast enough to be done directly. If not, pump it to
some local database, gbak it, zip it, and send it to destiny, gbak it back.
b) Create a small application to pump your blobs to files. The file name
will have the ID of the blob on the row. Or imagine something if rows can
have multiple blobs. Just identify the place at filename level, or you can
use another file to keep track of relation between ID-FieldName-FileName.
c) Copy this files to destin. Ftp, CDROM, whatever. You have the files now,
so you can zip them and test around.
d) Create a small application that will pump it again into the database. Run
it on destin.
In the future, you don't need a two way replication to do this.
If you have windows servers on both sides, you can easily build a one way
replicator with this two utilities and some version control inside your
database. (If you have Linux servers just wait for the IBO version in Kylix
and you are in!)
I hope this helps.
Artur
With a slow connection, transfer 69gigs will be something that I will never
think about. :-)
Even with ZeBedde. If we are talking about images, sometimes the image
format is already compressed, so....
Here is a possible solution.
You will need to have some work, but this is an easy task with Delphi+IBO:
a) Pump the initial database as you are already doing, without the BLOB
data.
This transfer will be fast enough to be done directly. If not, pump it to
some local database, gbak it, zip it, and send it to destiny, gbak it back.
b) Create a small application to pump your blobs to files. The file name
will have the ID of the blob on the row. Or imagine something if rows can
have multiple blobs. Just identify the place at filename level, or you can
use another file to keep track of relation between ID-FieldName-FileName.
c) Copy this files to destin. Ftp, CDROM, whatever. You have the files now,
so you can zip them and test around.
d) Create a small application that will pump it again into the database. Run
it on destin.
In the future, you don't need a two way replication to do this.
If you have windows servers on both sides, you can easily build a one way
replicator with this two utilities and some version control inside your
database. (If you have Linux servers just wait for the IBO version in Kylix
and you are in!)
I hope this helps.
Artur