Protecting Oracle Databases with HCC Using Veritas NetBackup
As organizations continue to adopt Oracle Hybrid Columnar Compression (HCC) to achieve substantial data reduction and performance optimization for data warehousing workloads, it becomes increasingly critical to ensure robust protection strategies for these databases. Veritas NetBackup delivers comprehensive support for Oracle databases leveraging HCC, seamlessly integrating with Oracle Recovery Manager (RMAN) to enable high-performance, policy-driven backups that maintain data integrity across both HCC-enabled and non-HCC storage environments. NetBackup leverages RMAN’s block-level capabilities to perform efficient full and incremental backups of Oracle databases, including those utilizing HCC on Oracle-engineered systems like Exadata and ZFS Storage Appliances. Through tight RMAN integration, NetBackup preserves the compression format during backup, minimizing storage consumption and backup windows while supporting deduplication for further optimization. This allows database administrators to maintain their HCC benefits even in backup and recovery workflows, while centralizing control and reporting through NetBackup’s interface. In recovery scenarios, NetBackup supports full flexibility whether restoring to an HCC-capable system or to a non-HCC environment. When recovering to an HCC-compatible platform, the compressed data remains usable in its native format, preserving both performance and storage efficiency. If recovery is directed to a non-HCC destination—such as traditional file systems or third-party storage—the restore process completes successfully, but the data must be uncompressed before it can be accessed. Full guidance on these options is available in Oracle’s official documentation on restoring HCC tables. NetBackup ensures that backup and recovery operations remain consistent and reliable across both HCC and non-HCC environments.
Oracle RMAN incremental to MSDP very slow
We are running RMAN backups to MSDP. The fulls and archive backups are very fast and dedupicate pretty well using filesperset 1. When the incremental backups run, it can take as much as 9 minutes to back up a 288k file, which makes no sense. It looks like bptm is waiting on something, but I cannot understand what it could be. Oracle is 10G, changed block tracking is enabled. The DBAs set filesperst to a higher number, and its faster, but it hardly deduplicates, so that's not gong to work. Anyone else see anything like this? I don't think its due to deduplication - it seems to be something on the oracle side. All the other jobs that run against the MSDP run very well. The behavior is the same using client side dedupe and server side dedupe. Master and Media are on 7.5.0.4 Oracle boxes are on 7.1.0.4 due to an old kernel they run.
Slow backup to tape
Hi, We're having some trouble backing up virtual and physical servers to tape. We're not getting the speed it should get and it takes up a lot of time to back up the environment. 2,63TB of data takes up more than 24 hours. Information about job, tape & tape drive: Tape type: HPE LTO Ultrium 6 6,25TB MP RW Tape Drive: MSL 2024 Compression: Hardware Encryption: None Verify: Yes These backups are going through the internal network within the domain. The average backup speed is 4,300MB/m (which is way to slow) When we backup locally on the backup server (for example the D:\ drive where the incremental disk backups are saved on the backup server) it gets to a speed of 10500MB/m. Which is understandable considering it's a straight connection to the tape drive. What did we do to try and resolve it: * Checked the compression * Checked encryption * Replaced tapes for new tapes * Resetted the tape drive * Updated Tapedrive firmware * Checked out the network for inconsistencies (not found), file transfers via other apps around 126mbs from server to server) * Job from VM on hyper-visor to tape (around 2,900 MB/m) * Local disk to tape (between 10500MB/M and 13500MB/M) * VM to disk on the backup server (Around 6200MB/M) Help would be appreciated.
