Scheduling and incremental backup in DSA

Backup Archive and Restore (BAR)
All things BAR
Highlighted
Enthusiast

Scheduling and incremental backup in DSA

Hi All,

Your thoughts on the below points please..

 

1. Do we have scheduling option in DSA through command line rather than below way. 

Creating the script with run_job command and scheduling that script through crontab or through other schedulers.

2. As per my understanding, DSA will identify changed records basing on the datablocks which got change from previous backup. That means there is no need to specify the column basing on which it has to track the changes of a table and there is no advantage even if table is having PPI.

=======================

Incremental backup is appropriate for:
• Databases and tables that have a very low change rate compared to table size
• Primary Partition Index (PPI) tables for which changes are limited to one or few
partitions

======================= 

14 REPLIES
Teradata Employee

Re: Scheduling and incremental backup in DSA

Re 2: It very much makes a difference to have PPI. When PPI is on an attribute that is correlated to the loading, then many fewer blocks in the table will be changed and therefore many fewer blocks will need to be backed up in the incremental. Without PPI, a medium to large load job could affect most or all blocks in the table and result in all the blocks needing to be backed up. (Note that incrementals do not back up changed records, they back up changed blocks. So if a single record or all of the records change in a block, the block will be backed up.)

Junior Supporter

Re: Scheduling and incremental backup in DSA

Currently DSC commandline and BAR protlets do not have scheduling options.
Apart from the 2 options that you mentioned if you are using Netbackup then you may want to check if you can schedule policies
and set them to run at specific time.


You do not have to specify any column for DSA to identify changed blocks.
Actually its the data deduplication solution (like data domain) which identifies repeat & new data patterns.


Abhishek Jadhav
Enthusiast

Re: Scheduling and incremental backup in DSA

Thanks Todd and Abhishek.

 

I am trying to understand how internally identifies the changed data blocks for cumulative and delta backups. Appreciate if someone can throw some light on this.

 

 

Approach 1:

 Day 1: Full backup

 Day 2: Cumulative backup: It will capture all the datablocks that changed after Day 1 full backup.

 Day 3: Cumulative backup: Will capture all the datablocks that changed from Day 1 full backup. Ideally it may need not to refer Day 2 backup image as any how it has to capture all the changes from Day1 full backup. So it may compare Day1 full backup image with the present database blocks and get the modified data blocks.

 

My question, is cumulative backup can always directly refer the last full backup and from there can get all the changed data blocks. 

 

Approach 2:

Day 1: Full backup. For example assume 10 data blocks.

Day 2: Delta backup: Will capture the changed data blocks after Day1 full backup. Blocks 7 and 8 changed from Day1 full backup so it will capture 7 and 8 data blocks. 

Day 3: Delta backup: Will capture the changed data blocks after Day 2 delta backup. Assume blocks 5,6 and 7 got changed.

 

My question: For delta backup, backup job has to consolidate the latest data block it is having from the full backup to till previous delta backup and then it has to compare to database blocks to identify the changed ones.

Solution: Before referring to database, Day 3 backup job, consolidates its latest data blocks as 1 to 6 from Day1 full backup, 7 and 8 from Day2 delta backup, 9 and 10 from Day1 full backup. This consolidated data blocks will be checked against actual database blocks.  

 

Teradata Employee

Re: Scheduling and incremental backup in DSA

The timestamp of the last backup is saved. Each datablock has a last updated timestamp. When an incremental backup is performed, the backup captures any blocks that have been changed since the last backup.

 

When a restore is required, the full backup is restored first. Then incrementals are restored starting from the most recent and working back. If a block has already been restored from a later backup, it will be skipped from the earlier ones as the restore works backward through the multiple incrementals.

Enthusiast

Re: Scheduling and incremental backup in DSA

The timestamp of the last backup is saved. Each datablock has a last updated timestamp. When an incremental backup is performed, the backup captures any blocks that have been changed since the last backup.

 

<Adithya> I think above statement applicable for delta backups. Can you explain how it will be for cumulative. For example,

cumulative backup on top of full backup: It may be similar to Delta backup. Captures the changed data blocks from full backup.

cumulative backup on top of previous cumulative backup : ???

cumulative backup on top of previous delta backups: ???

 

Teradata Employee

Re: Scheduling and incremental backup in DSA

The difference between the two types of incremental backup is which timestamp is chosen as the filter:

  • Cumulative - since last full backup
  • Delta - since last backup of any kind (full or incremental)
Teradata Employee

Re: Scheduling and incremental backup in DSA

Cumulative is related to part of the original question. If the updates/inserts/deletes are not closely correlated with PPI partitions, and there is any amount of data change, then cumulative backups will fairly quickly have to backup all the blocks. Using incremental to backup all the blocks is less efficient than just doing full backups.

Enthusiast

Re: Scheduling and incremental backup in DSA

Our plan is:

1. Friday we will take full backup

2. Monday we will restore the full backup

3. Next friday will take incremental backup and then restore it.

 

So in this scenario, which incremental(delta/cumulative) is better? And when we are restoring incremental, will it first delete the existing data in the restoring table, and then it will restore full backup data and then then restore the incremental data captured as part of incremental backup?

 

Is there any option which will restore only the incremental data? I mean only incremental data restoration rather than restoring last full backup data + incremental data

Teradata Employee

Re: Scheduling and incremental backup in DSA

Is this restore to another system?

 

If so, this plan will not work yet. In a future update of TD 16.20, Teradata will allow incremental backups to be restored to a system on an ongoing basis. Today however, it is onlly allowed to restore the full backup and all relevant incrementals as a single operation.