General Info

{{toc}}

h1. Database Mirroring 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 + incremental)</code> database backup

# Use <code>(full + delta)</code> database backup

# Use <code>(full + incremental + 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/

The following sections will describe how the last approach actually works. Before that, let's do a compare:

!database_mirroring_approach.png!

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>

# 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 standby database, just execute this command:

<pre>$ db2 rollforward db prod_db complete</pre>

The main selling points 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.

h2. Test Tools