Summary
This article defines OSS Archive jobs and explains how they differ from NDMP backups of DPX Filer data.
Resolution
OSS Archive refers to a job-centric automation that sends the latest disk-based backup job to tape. Running a OSS Archive backup job will only back up the latest snapshot associated with its OSS job. The OSS Archive job will re-create and transfer the total equivalent data that existed on the client machine when the OSS backup was run.
An OSS Archive Example:
Say you define an OSS job that backs up your C: drive, D: drive, and E: drive Monday-Friday and you schedule a OSS Archive backup on Saturday. Assume that each hard drive contains about 10 GB of data, and approximately 10% of each drive changes every day.
On Monday through Friday, block-level incremental backup jobs between the client machine and the OSS server will transfer about 1 GB of data from each drive. Thus, the total data transfer for each weekday will be about 3 GB.
On Saturday, the OSS Server will back up Friday's snapshot of this block-level incremental job. The OSS Archive job will then send the snapshot's contents to tape. In this case, the job will transfer 30 GB of data to tape because the source server has 3 drives with 10 GB of data on each drive.
To visualize this, imagine that the OSS Archive request does an IA map of your drive and then backs up that data to tape. If you have ever tried an IA map from an OSS Server, you know that the IA map represents the entire state of the source drive at the time the backup was run, even if every incremental backup is very small. Each time a OSS Archive backup job runs it will recreate the source drive data and send the total amount of data to tape. The OSS Archive job will back up the contents of each drive on your server every time the job is run.
Comparing OSS Archive Jobs to OSSV and NDMP Jobs:
OSS Archive backup jobs significantly differ from OSSV and NDMP backup jobs and require proper planning.
Near the end of an OSSV backup job, the Filer takes a snapshot of the destination volume. Regardless of how many jobs are sending data to a single volume, each OSSV job will take snapshots of the volume in its entirety.
Typically, when an NDMP 'dump' backup job is created for a Filer, you will use a special notation to back up the latest snapshot. The Filer will then backup all of the data found in that snapshot, regardless of that data's source. The data in a Filer snapshot represents the total client source data at the time of backup. Remember, multiple servers could be contained in the snapshot as the result of the volume being used for multiple jobs.
When an NDMP 'smtape' backup job is created for a Filer, the entire volume must be backed up. If you use the Ontap Command Prompt or the FilerView GUI to review the total data used in the volume, all that data will be stored on tape. This can include all jobs and servers that sent a base backup to that volume as well as the sum of all incremental backup data stored on that volume.
No NDMP backup scheme is directly analogous to an OSS Archive backup scheme because an OSS Archive job's destination OSS Server does not take snapshots at the volume level. The end of the OSS Archive job solely takes snapshots of data that was transferred as part of the backup job. Thus, running a OSS Archive job only backs up the base and cumulative incremental data represented on the last snapshot found for that job. Essentially, a OSS Archive job recreates the source drive data at the time of the OSS backup and sends all that data to tape.
OSS Archive jobs differ from NDMP 'dump' jobs because OSS Archive jobs will not back up data that might have been part of other OSS jobs. With OSSV jobs, anything sent to the destination volume will be sent to tape. OSS Archive jobs only send data that represents the original job's source drives to tape.
No direct functional equivalent exists between running a NDMP 'smtape' job and running an OSS Archive job. However, similar results can be accomplished by either running an Image backup of your OSS Server windows volume, or running an OSS backup your OSS Server drives to an alternate OSS Server.
The frequency that you schedule OSS Archive within an OSS job will depend on the local site's policies. Keep in mind that a OSS Archive backup only represents one OSS job client data backup.
Also, note that all OSSV and NDMP backup operations are accomplished in at least 2 steps:
- OSSV Backup of the client to the Filer in one job
- NDMP Backup of the volume in another job