evilution
ADSM.ORG Member
I'm trying to find out how mid sized companies are handling their database backups.
Currently, on each server, we provide the DBAs with a chunk of SAN disk where they spin off a copy off a flat file copy the database. After they complete their backup to disk they run a dsmc command to send that file to TSM.
This process works great as it provides them an on disk copy to use as they seem fit. For example they can rename and retain the file as a gold copy for as long as they want or they can use the existing file to perform prod-to-qual restores/refreshes whenever they want. In fact they have written scripts to do just that. They have automated the entire refresh process for qual and test. Access is controlled via NTFS perms and service accounts are used to process the restore request.
When building the servers we ask them how much storage they need for one full copy fo the database and that is what we give them. The trouble is sometimes they ask for more than really need. Back in the day we overlooked this behavior because storage wasn't as expensive and databases weren't as big.
Today we can no longer support the process of providing each server with SAN disk for backups. We are now entertaining using direct to TSM TDP backups for DR but when discussing this with the DBAs they questioned how they will support the refresh operations. They provided us with statistics that show APP/DEV is running in upwards of 50 restore operations each day. These are all using the flat files on disk and not TSM. Some of these are huge restores some are very small but long story short TSM would be unable to sustain this type of load. So we are kinda back to square one.
I'm just curious how everyone else supports not DR but database refresh operations, gold code redirected restores and production to test type operations. Please chime in with what you are doing.
Currently, on each server, we provide the DBAs with a chunk of SAN disk where they spin off a copy off a flat file copy the database. After they complete their backup to disk they run a dsmc command to send that file to TSM.
This process works great as it provides them an on disk copy to use as they seem fit. For example they can rename and retain the file as a gold copy for as long as they want or they can use the existing file to perform prod-to-qual restores/refreshes whenever they want. In fact they have written scripts to do just that. They have automated the entire refresh process for qual and test. Access is controlled via NTFS perms and service accounts are used to process the restore request.
When building the servers we ask them how much storage they need for one full copy fo the database and that is what we give them. The trouble is sometimes they ask for more than really need. Back in the day we overlooked this behavior because storage wasn't as expensive and databases weren't as big.
Today we can no longer support the process of providing each server with SAN disk for backups. We are now entertaining using direct to TSM TDP backups for DR but when discussing this with the DBAs they questioned how they will support the refresh operations. They provided us with statistics that show APP/DEV is running in upwards of 50 restore operations each day. These are all using the flat files on disk and not TSM. Some of these are huge restores some are very small but long story short TSM would be unable to sustain this type of load. So we are kinda back to square one.
I'm just curious how everyone else supports not DR but database refresh operations, gold code redirected restores and production to test type operations. Please chime in with what you are doing.
Last edited: