FileSlinger™ Backup Reminder 04-07-06: The Limits of Drive Imaging

The Ur-Guru, as I may have mentioned, practically has a data center in his home office. I’ve lost track of how many different computers he has, and the fact that he’s got multiple hard drives with virtual machines installed on most of them so he can test new software only makes it more complicated. He’s starting to run out of names for them, and he could create a wall-sized display with his network map. There are businesses with 20 employees who don’t have as many computers or as much data to back up.

If you have more computers than you can count, you need a different backup strategy than people who have only one or two machines to worry about. For years my 80 GB 2.5” external hard drive has been more than sufficient to hold two or three Ghost images of my laptop’s drive (30 GB for Keramat and 20 GB for her temporary replacement, Astarte), plus software installation packages and file backups of my current working files.

But now, as I said last week, I have so much storage I don’t know what to do: my new laptop, Enheduanna, with *two* 80 GB hard drives. If I filled them both up (which the “95% rule” of storage says I eventually will), there’s no way I can back up Enna’s internal drives to my 80 GB XHD.

(Oh, and in case you’re wondering why I name my computers: so I can tell them apart. It’s easier than saying “The Enpower,” “The Dell,” “The HP widescreen.” If you’re wondering why I name my computers after figures from Near Eastern mythology and history, it’s because I actually have a lot more expertise in ancient civilizations and languages than I do in computers. As for why I give them nicknames—because I pick names that are too long. ;-))

Whether it’s my own two-laptop home office or the Ur-Guru’s mind-boggling crowd of machines, however, increased internal storage capacities require a new approach. Naturally he and I started discussing it as soon as I knew I was getting a new computer. Since he’s the one who has expanded his drive capacity so many times that he’s turned file storage into an art form, I thought I’d let him speak for himself. I’m compressing several messages exchanged over the course of a week or so, with a few edits for clarity.

As the 160GB ‘fills up’ (to whatever degree) the Ghost images also increase by a lot making imaging a very inefficient system. For example, on my laptop I carry some demo data that I can copy back on it at any time, but the total load is like 4GB. If I were to have just one partition on the laptop every image would include that 4GB which is a waste of time and space. That’s why I chopped the HD into two partitions. One with mostly ‘static’ data that I need to have around, one system partition that I can backup and image as needed (which includes work-data so that gets to be part of the image).

So you should separate data and system on the two disks so that not all the same stuff goes into each new image. Especially since those images would grow larger over time and as the disks fill up. I have this system where the data is divided into different categories (they may live on the same set of disks instead of separate ones, depending on the system):

– Static
– Dynamic
– Work
– On Hold
– Server
– System

Static is everything that I want on a machine, usually on at least two or three machines, and is data that *never* changes at all except that from time to time stuff gets added to it.
For example, my Deploy share that contains all the stuff I need to install various machines, drivers, apps, etc. and that’s like 36GB (all archives and compressed, of course).

Every few months that set of data is updated with updates, new apps, etc. It makes no sense to make frequent backups or images so static data is simply synced and replicated to three machines so they all have the same stuff “handy”.

Dynamic data is like static data except it changes more frequently and individual files/folders/hierarchies will change more often. Dynamic data is also treated as static, it’s not part of any images and backups are done by archives and replication.

Work data is what I’m working on, current data, etc. This includes projects, source files, URLs, folders, the whole shebang. This is part of images, replication, and archiving and that data is backed up via a “paranoid” level system to ensure several backups exist of that data along with several backups back in time (about 1 month I can go back).
Work data is never backed up “as a whole”. Some stuff gets backed up every week, some stuff every day, some stuff gets done twice a day automatically. Work data often gets to be dynamic data over time (as projects no longer get updated or files need to be kept handy but are not likely to change too much. Dynamic data, however, often becomes static data; this includes projects and stuff that need to be kept but will certainly not change anymore (or is very unlikely to change).

On Hold stuff is files that are on hold until I decide what to do. This is replicated on two machines only and data either gets deleted over time or backed up to DVDs for “keeping”.

“Server” is data which is replicated between both servers so they can take over at any time when needed. This happens a few times a day, automatically, and once a day it gets collected and archived to another machine which acts as a “safe copy” of the server data, on a daily basis. These archives just keep adding and piling up until I delete the older ones manually.

And then of course there’s the system, which is simply imaged. My C: partitions are all 50GB which is sufficient for all machines. Images are anywhere between 9GB and 40GB depending on the system.

I have completely moved away from Ghost these days except for one machine. Now with [True Image 9’s] file level backups and incremental stuff I think I might turn it into my “main” backup app (that is, apart from all my custom scripts that do sync copies on the network). I can’t use Ghost on some of my systems and I don’t want to deal with Ghost images and True Image images, I’d want just one thing do “do ’em all”, so True Image fits better in that respect. Other than that I do prefer Ghost running outside of the OS which True Image does not.

I haven’t tried True Image yet, but may find myself using it in the future. In the meantime, I’m working toward an arrangement which will synchronize my working files between the C (system) and D (data) drives on Enna in addition to re-creating the system I had with Kera and Star, where I use Karen’s Replicator to copy my important documents and Outlook data onto the XHD every time I boot the machine and use Ghost for system snapshots.

As of this moment I have 14.4 GB free on my XHD and 57 GB filled on my C drive. That’s going to make storing a Ghost image on the X drive challenging, even with high compression. (As for copying the D drive—that’s what DVDs are for, until and unless I get one of those nice network attached storage devices. DVDs can be mailed elsewhere or stored in a safe deposit box, though.) But I already have the basic system installation and major applications saved in Ghost images, so it may be time to look at True Image and start going the incremental route.

I’ll let you know what happens.