Database Migration

IDS9.40.FC4 HPUX 11.0 Database size 200Gb Hello all, I am about to migrate a database to a larger server and was wondering which method is recommended. The easiest method I guess is to use a level 0 archive which would mean the same structure between the two servers which the Unix guys building the system are against. dbexport I fear would take too long, as would need to be done to tape (size restrictions of system) Which leaves onunload/onload....This is just a data unload and a schema of the whole database would need to be taken prior to onunload running (is this right?) Is this the same as dbexport where the chunk raw-logical-volumes names and sizes do not need to be as the original machine. Has anyone out there got an easy explanation of the process involved when using onunload/onload? Cheers, Howard

Why can't you do a redirected restore? also -- don't count out HPL. "Howard Jones" <howie_lfc@hotmail.com> wrote in message news:QMb6d.562$_V.161@newsfe6-gui.ntli.net... > IDS9.40.FC4 > HPUX 11.0 > Database size 200Gb > > Hello all, > > I am about to migrate a database to a larger server and was wondering which > method is recommended. > > The easiest method I guess is to use a level 0 archive which would mean the > same structure between the two servers which the Unix guys building the > system are against. > > dbexport I fear would take too long, as would need to be done to tape (size > restrictions of system) > > Which leaves onunload/onload....This is just a data unload and a schema of > the whole database would need to be taken prior to onunload running (is this > right?) > Is this the same as dbexport where the chunk raw-logical-volumes names and > sizes do not need to be as the original machine. > Has anyone out there got an easy explanation of the process involved when > using onunload/onload? > > Cheers, > > Howard > >

have you considered onpload? "Howard Jones" <howie_lfc@hotmail.com> wrote in message news:<QMb6d.562$_V.161@newsfe6-gui.ntli.net>... > IDS9.40.FC4 > HPUX 11.0 > Database size 200Gb > > Hello all, > > I am about to migrate a database to a larger server and was wondering which > method is recommended. > > The easiest method I guess is to use a level 0 archive which would mean the > same structure between the two servers which the Unix guys building the > system are against. > > dbexport I fear would take too long, as would need to be done to tape (size > restrictions of system) > > Which leaves onunload/onload....This is just a data unload and a schema of > the whole database would need to be taken prior to onunload running (is this > right?) > Is this the same as dbexport where the chunk raw-logical-volumes names and > sizes do not need to be as the original machine. > Has anyone out there got an easy explanation of the process involved when > using onunload/onload? > > Cheers, > > Howard

"Howard Jones" <howie_lfc@hotmail.com> wrote in message news:QMb6d.562$_V.161@newsfe6-gui.ntli.net... > IDS9.40.FC4 > HPUX 11.0 > Database size 200Gb > > Hello all, > > I am about to migrate a database to a larger server and was wondering which > method is recommended. > > The easiest method I guess is to use a level 0 archive which would mean the > same structure between the two servers which the Unix guys building the > system are against. In IDS 9.40 you have the option of a re-directed restore, where you can alter the chunk names at restore time. Whilst this still requires you to have the same number and sizes of chunks as per the original server, this ought not to be too restrictive. If your old and new server can be connected to the same SAN (not necessarily at the same time), you could do quite a neat, and very quick, cutover using Informix mirroring. I notice that a couple of other (and very switched-on) correspondents suggest HPL. Unless you have a small number of tables I personally would advise against this though. Onunload and unload might work, but I think might be pretty slow, and you'd need to test it first, perhaps on an empty database created from the schema of your live one, as it's pretty picky on dbspace names etc. (I have it in my mind that onload used to be as fast as an ontape restore, but now is much, much slower. Does anyone else share this recollection of an ageing mind?).

On Tue, 28 Sep 2004 11:19:12 GMT, "Howard Jones" <howie_lfc@hotmail.com> wrote: >IDS9.40.FC4 >HPUX 11.0 >Database size 200Gb > >Hello all, > >I am about to migrate a database to a larger server and was wondering which >method is recommended. > >The easiest method I guess is to use a level 0 archive which would mean the >same structure between the two servers which the Unix guys building the >system are against. > >dbexport I fear would take too long, as would need to be done to tape (size >restrictions of system) > >Which leaves onunload/onload....This is just a data unload and a schema of >the whole database would need to be taken prior to onunload running (is this >right?) >Is this the same as dbexport where the chunk raw-logical-volumes names and >sizes do not need to be as the original machine. >Has anyone out there got an easy explanation of the process involved when >using onunload/onload? > Try HPL . . . the onpladm utility allows command-line HPL job creation and execution . . . It would be sweet . . . JWC