Database Mirroring (DB2) on remote site » History » Revision 12
Revision 11 (chin-yeh, 03/22/2012 10:22 AM) → Revision 12/23 (chin-yeh, 03/22/2012 10:33 AM)
{{toc}}
h1. Database Mirroring (DB2) on remote site
h2. Introduction
To mirror production database(s) to remote site, there are a number of approaches:
# Use <code>(full)</code> database backup
# Use <code>(full + cumulative)</code> database backup
# Use <code>(full + delta)</code> database backup
# Use <code>(full + cumulative+ delta)</code> database backup - _not recommended because it is too complex_
# Use <code>(full database backup + transaction logs)</code>
> the approach 1-4 are described in "Using DB2 incremental backup":http://www.ibm.com/developerworks/data/library/techarticle/dm-0910db2incrementalbackup/
We will only discuss about the last approach, _full + transaction logs_ in the coming sections. Before that, let's do a quick compare for these approaches:
||_.full|_.full + cumulative|_.full + delta|_.full + cumulative + delta|_.full + transaction logs|
|*Backup Image Size*|Biggest|Small|Smallest|In between Small and Smallest|Smaller|
|*Backup Time*|Longest|Depends on how the data is stored across tablespaces. If all data is stored in one tablespace then it won't make any difference|Depends on how the data is stored across tablespaces. If all data is stored in one tablespace then it won't make any difference|Depends on how the data is stored across tablespaces. If all data is stored in one tablespace then it won't make any difference|Fast|
|*Easy To Recover*|Easiest|Easy|Difficult|Very Difficult|Easy|
|*Need To Revise Backup Strategies*|No|Yes|Yes|Yes|No|
|*Resources Utilization*|High|High to Medium|High to Medium|High to Medium|Low|
|*DB is usable after first restored*|Yes|No|No|No|No|
|*Functions Used*|backup + restore + rollforward|backup + restore + rollforward|backup + restore + rollforward|backup + restore + rollforward|backup + restore + rollforward + rsync|
h2. Full Database Backup + Transaction Logs
Below diagram describes how this approach works:
!full_transaction_logs_approach.png!
_Sample steps:_
# Like any other approaches, you must perform a full database backup _(either offline or online)_
** e.g.:
<pre>$ db2 backup db prod_db online</pre>
# Transfer the full backup image to the standby site and perform a database restoration:
** e.g.:
<pre>$ db2 restore db prod_db replace history file</pre>
# Use the <code>rsync</code> command to do an one way incremental sync from remote (production) site to standby site.
** e.g.:
<pre>$ rsync -var --delete db2inst1@db-prod:/home/db2inst1/db2inst1/NODE0000/SQL00001/SQLOGDIR/ /home/db2inst1/db2inst1/NODE0000/SQL00001/SQLOGDIR</pre>
# Execute the <code>rollfoward</code> on the standby database: (*do not* include the *complete* option)
** e.g.:
<pre>$ db2 rollforward db prod_db to end of logs</pre>
# If want to bring up the standby database, just execute this command (_change the database alias_):
<pre>$ db2 rollforward db prod_db complete</pre>
The biggest advantage of this approach is there's little or no need to revise current backup strategies, however, the mirror database cannot be used until the <code>rollforward</code> process is terminated (_also happens on the other db incremental backup approaches_).
h2. Test Tools
To test run those database backup