Outsourcing of backups to the Oracle Cloud to ensure geo-redundancy

Summary

From a security perspective and to ensure geo-redundancy, it is increasingly important not only to have backups available locally on disk, but also to back them up separately from one another. For this purpose, the Oracle Cloud solution is ideal for outsourcing backups for warranty purposes of increased redundancy and geo-redundancy.

Business challenge/problem that was resolved

For our customer Hugendubel, it is important not only to have backups available locally on disk, but also to prevent failures due to catastrophes, storms and accidents should they occur. The spatial distance makes sense here to protect against failure or loss of data information. The Oracle Cloud solution is ideal for outsourcing backups to ensure the Backup redundancy and ensure geo-redundancy.
In addition, during implementation and evaluation, analyzes were carried out with regard to parallelism and compression in order to save storage space and costs.

The backups are currently carried out according to the following scheme: A level 0 backup (full backup) is carried out every week on Sunday. This results in a complete backup of all data regarding the databases. This is a consistent copy of the database that can be restored on another server. In addition, a level 1 backup (incremental backup) is performed daily for this data, which means that only the changes for the current day are backed up and extended with the backup of the previous day.
The backups are currently written to Oracle ZS3 hard disk drives (HDD) in the supercluster. These ZS3 drives are mirrored to further minimize the risk of data loss.
Each of these backups contains both a physical (backup) and a logical backup (export) of the data. A logical backup does not save any physical data blocks, but exports the metadata of the tables and their contents. A physical backup, on the other hand, allows all transactions to be restored. The backups to the cloud should be carried out in combination with the local backup. The rhythm between level 0 and level 1 backups should be taken from the actual state.

For the implementation, a separate storage bucket was created in the Oracle Cloud for each database, and a separate IAM user was also created for each database.
After the API keys were created, the Oracle Cloud Backup module was installed separately for each database. After that, automatic archiving was switched on and the standard channel (SBT_TAPE) was configured for backup in RMAN.
To improve the upload speed, parallelism tests were made. For this purpose, the parallelism was adjusted taking into account the assigned vcpus of the respective vm semultan with the parallelism of the local backups and thus a significant improvement in upload speed was achieved.
In order to reduce the backup set sizes and costing for object storage, the compression setting prescribed by Oracle was changed to medium. We were able to achieve storage space savings and cost savings of 47%.

Why Oracle

Our customer has been relying on the software and structures available from Oracle for years. In addition, migrating existing enterprise workloads improves reliability and performance for all applications, and provides the complete services customers need to develop innovative applications. In addition, there is also the first-class price-performance ratio.

Results

The Oracle Cloud solution provided by us with regard to the geo-redundant protection of the backups and their outsourcing to the cloud, a further security aspect is fulfilled and thus offers an important aspect in the event of a disaster.

In terms of upload speed, we have achieved a reduction in incremental backup sets from 9 hours to around 3 hours, and in incremental level 0 backups there has been a reduction from 63 hours to around 16 hours.
To reduce the backup set sizes and costing for object storage, the compression setting prescribed by Oracle was changed to medium. We were able to achieve storage space savings and cost savings of 47%.

And to monitor the backup process, a metric-based monitoring took place in the Enterprise Manager, which clearly displays the backup status of the individual databases.