Copying a database - suggestions please

IDS v7.24. UC6, HP-UX 10.20. I need to make a copy of a large dataabse (about 30GBytes) from one database server to another but on the same UNIX server. There are the options I've considered: 1. onunload/onload - a non-starter as a big in onload means that IDS tries to put detacted indexes back into their original dbspaces - but there's no equivalent option to put the tables back into original spaces. 2. ontape back-up/restore - will only work if I do something smart-arse with chroot 3. DBMS or hardware mirroring - not possible to make any use of this once the links are broken. So I'm left with good old, slow old, dbexport/dbimport. Unless anyone has a better idea???????? thanks Neil

How about high performance loads using pipes? This requires no disk space, hence no I/O. Since the two servers are on the same machine, no network contentions. Good luck, Arun Neil Truby wrote: > IDS v7.24. UC6, HP-UX 10.20. > > I need to make a copy of a large dataabse (about 30GBytes) from one database > server to another but on the same UNIX server. There are the options I've > considered: > > 1. onunload/onload - a non-starter as a big in onload means that IDS > tries to put detacted indexes back into their original dbspaces - but > there's no equivalent option to put the tables back into original spaces. > > 2. ontape back-up/restore - will only work if I do something smart-arse > with chroot > > 3. DBMS or hardware mirroring - not possible to make any use of this once > the links are broken. > > So I'm left with good old, slow old, dbexport/dbimport. Unless anyone has a > better idea???????? > > thanks > Neil

Got a script for doing all tables in a db? And how would it compare to dbexport/dbimport do you think? Thanks Neil Arun Shastry wrote in message <36FA8A8A.2B9448C5@accentsoftware.com>... >How about high performance loads using pipes? > >This requires no disk space, hence no I/O. Since the two servers >are on the same machine, no network contentions. > >Good luck, >Arun > >Neil Truby wrote: > >> IDS v7.24. UC6, HP-UX 10.20. >> >> I need to make a copy of a large dataabse (about 30GBytes) from one database >> server to another but on the same UNIX server. There are the options I've >> considered: >> >> 1. onunload/onload - a non-starter as a big in onload means that IDS >> tries to put detacted indexes back into their original dbspaces - but >> there's no equivalent option to put the tables back into original spaces. >> >> 2. ontape back-up/restore - will only work if I do something smart-arse >> with chroot >> >> 3. DBMS or hardware mirroring - not possible to make any use of this once >> the links are broken. >> >> So I'm left with good old, slow old, dbexport/dbimport. Unless anyone has a >> better idea???????? >> >> thanks >> Neil >

Ok...we're now talking pro's and cons. The high performance loader approach needs some amount of setup before you actually run the jobs. It would take more time if you use their UI, check the onpload database and start inserting records into the tables there. You probably know that the HP load uses this database to understand the unload/ load jobs. This approach is far better and much, much faster than the dbexport/dbimport. However, dbexport/import needs zero setup time. From my experience as an Informix DBA, I've used HP loader with pipes for big databases (>100GB), but with fewer tables, and dbexport/import for smaller databases with lots of tables. I've also used a combination of the two sometimes - HP load big tables (remember dbexport will bombout if one of your tables exceeds the 2GB limitation on a 32 bit hardware/OS), and simple unload SQL scripts for smaller tables. If you want a script, check Mark Scranton's website on tripod.com. If you still can't find them, I could dig'em up from my archives (I've stopped informix work for a while now!). Good luck, Arun Neil Truby wrote: > Got a script for doing all tables in a db? And how would it compare to > dbexport/dbimport do you think? > > Thanks > Neil > > Arun Shastry wrote in message <36FA8A8A.2B9448C5@accentsoftware.com>... > >How about high performance loads using pipes? > > > >This requires no disk space, hence no I/O. Since the two servers > >are on the same machine, no network contentions. > > > >Good luck, > >Arun > > > >Neil Truby wrote: > > > >> IDS v7.24. UC6, HP-UX 10.20. > >> > >> I need to make a copy of a large dataabse (about 30GBytes) from one > database > >> server to another but on the same UNIX server. There are the options > I've > >> considered: > >> > >> 1. onunload/onload - a non-starter as a big in onload means that IDS > >> tries to put detacted indexes back into their original dbspaces - but > >> there's no equivalent option to put the tables back into original spaces. > >> > >> 2. ontape back-up/restore - will only work if I do something > smart-arse > >> with chroot > >> > >> 3. DBMS or hardware mirroring - not possible to make any use of this > once > >> the links are broken. > >> > >> So I'm left with good old, slow old, dbexport/dbimport. Unless anyone > has a > >> better idea???????? > >> > >> thanks > >> Neil > >

Ever consider to buy a RAID storage system? You'll never need to do any backup once you get one. Neil Truby wrote in message <7ddli4$227$1@taliesin.netcom.net.uk>... >IDS v7.24. UC6, HP-UX 10.20. > >I need to make a copy of a large dataabse (about 30GBytes) from one database >server to another but on the same UNIX server. There are the options I've >considered: > >1. onunload/onload - a non-starter as a big in onload means that IDS >tries to put detacted indexes back into their original dbspaces - but >there's no equivalent option to put the tables back into original spaces. > >2. ontape back-up/restore - will only work if I do something smart-arse >with chroot > >3. DBMS or hardware mirroring - not possible to make any use of this once >the links are broken. > >So I'm left with good old, slow old, dbexport/dbimport. Unless anyone has a >better idea???????? > >thanks >Neil > >

Baloney. You always need to do backups. What if a power surge takes out the array? Sounds like you've been listening to EMC's sales talk. (As Girsch prepares to be flamed by EMC and its loyalits...) X^p Ye Chenwen wrote in message <7dspl1$9rk$1@nobel2.pacific.net.sg>... >Ever consider to buy a RAID storage system? You'll never need to do any >backup once you get one. > >Neil Truby wrote in message <7ddli4$227$1@taliesin.netcom.net.uk>... >>IDS v7.24. UC6, HP-UX 10.20. >> >>I need to make a copy of a large dataabse (about 30GBytes) from one >database >>server to another but on the same UNIX server. There are the options I've >>considered: >> >>1. onunload/onload - a non-starter as a big in onload means that IDS >>tries to put detacted indexes back into their original dbspaces - but >>there's no equivalent option to put the tables back into original spaces. >> >>2. ontape back-up/restore - will only work if I do something smart-arse >>with chroot >> >>3. DBMS or hardware mirroring - not possible to make any use of this >once >>the links are broken. >> >>So I'm left with good old, slow old, dbexport/dbimport. Unless anyone has >a >>better idea???????? >> >>thanks >>Neil >> >> > >

Bollocks! You've been listening to the guys in marketing too much. In any case, what if you had to restore because of an application error? Neil Ye Chenwen wrote in message <7dspl1$9rk$1@nobel2.pacific.net.sg>... >Ever consider to buy a RAID storage system? You'll never need to do any >backup once you get one.