Question

DB Full backup time overlap with Incremental backup time

  • 10 August 2023
  • 5 replies
  • 176 views

Badge +1

Hi All,

My oracle DB Size is 40TB and scheduled as:
Full: Sunday -4PM
Inc: Daily -4PM

1) Now, I am facing an issue for run full backup because is take more 24Hours to complete the job.

     -What impact on DB or machine itself if Full Job overlap with Incremental ?
     -Its is ok for both job ( Full and Incremental) run overlap ?
2) Now, full backup already take 25 hrs for complete the job, what estimation time for restore job         complete? Any rule or formula to calculate time?

Hope any suggestion from experience person.


5 replies

Userlevel 1
Badge +4

Hello @ghani 

Thanks for your query.

Both the jobs should not overlap. 

Usually it is presumed that restore will take more time than the backup job, thumb rule is, restore can take up to twice the backup time.

Also, I would recommend to refer following documentation for best practices and performance tuning.

https://documentation.commvault.com/11.24/expert/20811_best_practices_for_oracle.html

https://documentation.commvault.com/11.24/expert/20823_oracle_agent_general_agent_faqs.html

- Also the RMAN TUNING to speed up the backups to max:


https://docs.oracle.com/cd/B28359_01/backup.111/b28270/rcmtunin.htm#BRADV89857

 

Userlevel 4
Badge +10

Hi ghani,

 

Although the schedules might be overlapping due to the runtime of the Full Backups, both jobs should not run together.

 

The Full Backup will trigger on Sunday @ 4pm and will keep on running until completion

The incremental will trigger next day @ 4pm, if the previous Full Backup is still running, the job will queue until the Full Backup has finished, then it will start its operations.

 

This is completely normal and the application will handle it

https://documentation.commvault.com/2022e/expert/6385_job_status_descriptions.html

 

 

Badge +1

Hi @Tanuj Balani ,

 thanks for idea and thumb rule...but for RMAN tuning to speed up backup job i think not possible because currently the machine also have the performance issue.

Badge +1

Hi @Javier ,

thanks for the explanation. very appreciated 

Userlevel 4
Badge +13

Hi @ghani 

When a full backup is in progress, any job initiated for the same sub-client will fail unless job queueing is enabled. 

Regarding performance issue, the best way it to first identify the bottleneck and tweak the settings based on the findings.

  1. From the job details of the full job go to the details tab and verify the throughput which shows the break down of oracle read speeds, media agent and DDB speed. (paste the screenshot on this thread)
  2. Navigate to the oracle server and review ORASBT.log for keyword sbtps (cat ORASBT.log | grep -i sbtps | grep <jobID>    This will show Oracle I/O vs MA I/O and helpful to isolate and identify the bottleneck 
  3. Review the CvPerfMgr.log on the media agent for detailed performance counters 

Based on what you find, we could tweak the parameters  

 

Additionally could you answer the following

  1. Is this a TDE database ?
  2. Is compression or encryption enabled on the RMAN level ?
  3. How many streams are used for the backup
  4. If its using a MA on the network then whats the bandwidth between oracle server and MA, You may try checking a data copy test or (iperf utility) to verify the network bandwidth between oracle server and the media agent

Regards,

Gowri Shankar 

 

Reply