Data Backup and Recovery

Backup Oracle Daily-Full/Hourly Archive Suggestions (and Feature Request)

f_duranti
3,097 Views

We use a backup strategy in which we do a Full Oracle online backup daily (in which we also delete old archive logs) and hourly Archive log only backup.

We use Oncommand to do the mirroring.

To do this we need to have 2 different configuration file that, at this moment, differs only for the volume List.

The 2 configuration file have mostly the same parameters (oracle and archive):

CNTL_FILE_BACKUP_DIR=/saparch/GRC/controlfile_bck

ORACLE_BACKUPMODE=online

SQLPLUS_CMD=/oracle/GRC/112_64/bin/sqlplus

ARCHIVE_LOG_ONLY=daily:N;hourly:Y

ORACLE_HOME=/oracle/GRC/112_64

ORACLE_DATABASES=GRC:oragrc

ORA_TEMP=/tmp

ORACLE_EXPORT_PARAMETERS=N

ARCHIVE_LOG_RECURSIVE_SEARCH=N

ARCHIVE_LOG_DIR=/saparch/GRC

ARCHIVE_LOG_EXT=dbf

ARCHIVE_LOG_RETENTION=1.0

ARCHIVE_LOG_ENABLE=daily:Y;hourly:N

The only different thing is related to Volumes and Dataset.

For the full online we have:

VOLUMES=itromfs22:GRC_Log,GRC_Exe,GRC_Report,GRC_Arch,GRC_Data

META_DATA_VOLUME=itromfs22:GRC_Arch

NTAP_DFM_DATA_SET=itromfs22:snapcreator_GRC/GRC_Data,GRC_Exe,GRC_Log,GRC_Report;itromfs22:snapcreator_GRC_arch/GRC_Arch

For the archive log only we have:

VOLUMES=itromfs22:GRC_Arch

NTAP_DFM_DATA_SET=itromfs22:snapcreator_GRC_arch/GRC_Arch

I'm not sure if this is the correct way to do this so I'm asking for suggestion if anyone know some trick to simplify our configuration.

What came to my mind is that it could be nice if in the case of Oracle, at least, the volume snapshot/mirror/backup could be done in different way from SnapCreator Framework checking on the value of ARCHIVE_LOG_ONLY parameter.

I'll try to explain. Normally I think that you will have all volume in VOLUMES and the Archive Log Volume also in the META_DATA_VOLUME definition.

When doing an ARCHIVE_LOG_ONLY=N backup (we have daily:N) SC4.1 should work as usual:

1) Put oracle in backupmode

2) backup database (all volume excluding Archive-Metadata)

3) end backup

4) backup archive-Metadata

When SC find ARCHIVE_LOG_ONLY=Y it would be nice if will do the backup this way.

1) create archivelog

2) backup archivelog-metadata ONLY (instead of backing up the entire database)

Doing this without altering the actual configuration could be done putting a new option like M in addiction to Y/N to backup only metadata volumes instead of all volumes.

It would also be nice to be able to configure a "different" dataset name (like snapcreator_config_M or snapcreator_config_A) when creating the configuration from the GUI so that it will automatically create 2 different dataset.

Having 2 different dataset is important if using dfm because with just one dataset it will snapmirror the entire dataset and not just 1 volume.

1 REPLY 1

sivar
3,097 Views

Thank you Francesco. This is a Brilliant Idea!

I have created two Rally User Stories to our Engineering team to achieve the same.

US2055 - Requesting ARCHIVE_LOG_ONLY=M and associated workflow

US2056 - Ability to create 2 different datasets for a given config based on VOLUMES and META_DATA_VOLUMES

The product management team will weigh-in the options and address this for a future release.

Thanks,

Siva Ramanathan

Public