How to make a good storage choice if moving my confluence prototype installation a few times.

[Tim Penner]

I'm planning to move my prototype confluence installation at least a couple times, from a basic PC, to a powerful PC, and then (I'm sorta guessing here) to a Linux server, and then maybe an enterprise server. It occurs to me to use my fav opensource database (MySQL) as the storage mechanism so I can just unload/reload data at each step. Does this make sense? Am I over-complicating things? Should just trust Atlassian migration tools and take the path of least resistance with each installation? For example: just use the built-in database for my PCs and wait to use a database with the Linux server.

[Tim Penner · Accepted solution]

It turns out that importing from a backup is pretty "safe". I used it to convert my little installation from a built-in database to an external Postgresql database.

We apparently hung confluence by doing a huge nasty import (from word) running on the internal database, so I started to get nervous. We abandoned that little database and started again since we had only done some learning experiments. I restared quickly with the built-in base again and then, after a couple weeks of getting some nice test content looking decent, I created the new install with Postgresql and ported the data over. It worked perfectly. And the benefit (besides the comfort that I have a proper database running) is that, even with the little system I'm hosting this on, the whole thing seems to run a lot faster.

For those who might be curious about my "little installation", I have an aging 4-core (8-thread) I7 portable computer with 8 GB member running Windows 7 pro. Confluence is running on an Ubuntu 12.04 guest of Virtual Box. I gave it 4GB of the memory and 4 of the processors. Confluence is very snappy in this environment - kudo's to the performance team.

The ubuntu/confluence installation with postgresql was a piece of cake (compared to MySQL, at least). Using a VM means that if someone bequeathes me better hardware (and if I get money for more licences), there's no rebuilding. I just run-up Virtual Box again on any kind of monster host and copy the Ubuntu image to it. My backups are being sent to a large external drive managed by Windows.

[Martin Jæger]

My initial thought is that you might be overcomplicating. You would have to do some reversing of the database, in order to make sure that you extract what you need. I could also imagine that the import/export-feature has special 'treatment' around references between pages/spaces etc. But be aware when using the import/export. If you do a 'full' export (evertything), and import it into another instance, you will wipe ALL other content already there. Great power, great responsibility!