Sunday, August 5, 2012

Sticking it to "The Man"

I found myself shocked and surprised when I saw a line on a document the other day about possibly needing to maintain 2 images. This was from people in the know. So to all of those out there reading my blog with rapt attention about "The Image" - if you're not quite getting it, don't worry. Those in the project don't seem to either.

"The Image" (I really need to come up with a name for it) is really incredibly flexible. That stuff in Manaiakalani Shininess Part II - it's real. I mean, really real. Combine it with the stuff about supplements (packages at the moment but I've every intention of extending it to scripts) and there's some REAL flexibility.

For example, to add another site, it is no longer at all necessary to create a new image. Make the packages, throw them into the right folder on the KttC and the site is added. For a simple site, you're looking at less than a week to make up all of the branding packages, get network settings in there, add bookmarks etc.

That's KttC. Need to make some serious customisations? That site definition made while building the packages? Add software. Remove software. Do what ever you please. It's just a text file. I'm now making up a demo for installing KDE and removing Gnome (and the Netbook Interface). Now I'm not saying there isn't a sacrifice here. The bigger the changes, the more time it takes to set up the system when the user goes through their initial login process. That's a pretty big sacrifice - it effects the user's experience BUT it's perfectly possible (read: It may be necessary to maintain multiple images IF there are enough sites with enough big differences to justify it. It would have to justify the maintenance overheads).

The blueprints - it solves a whole other problem. Currently we're imaging. Imaging is fine - it's good. BUT it doesn't really give us any flexibility in partitioning. Look at any imaging system and you'll find that you can image to machines with bigger hard drives, but not smaller (there's absolutely no reason why this same image couldn't be used on a 40GB hard drive BUT the image has been made on a 160GB machine). Resizing partitions is also problematic - they attempt to do it proportionally rather than keeping things sensible (is there any real need to make a system partition bigger if it works just because you've got a bigger hard drive? Shouldn't the space rather be used for user generated files?). So part of this process will be to define maximum and minimum sizes for partitions.

The sacrifice? There's always some sort of sacrifice. The time needed to format a file system for each partition. Given that you no longer need a stick to do this though (except for the initial load of the rescue partition), this is probably less of a concern.

So "The Image" is good currently. It'll be even better soon. So to all of those doubters out there... Stick it.

No comments:

Post a Comment