In the On-Site And Off-Site Backups thread I mentioned that I had tried Carbonite and found that it did not work for me in my particular situation. Loki asked me to explain so I promised a thread dedicated to that subject. This is that thread ...

(Aside: I can see now that much of this material is going to want to be re-posted to the PC Software Tech forum. Oh well.)

xxxxxxxxxxxxxxxxxxxx

loki,

I have nothing against Carbonite in principle. I think it's a great idea, and I very much like its human interface, which is simple enough for most computer users to be able to use with no difficulty. In fact, since most people don't do backups at all, much less off-site backups, services like Carbonite are perfect because once the system is set up, the average user need not pay any attention to it.

xxxxxxxxxxxxxxxxxxxx

Folks,

Here's how Carbonite works ... (Or at least how it worked when I evaluated it in February of 2007).

You tell Carbonite what parts of your system you want backed up. You do this via checkboxes in a treeview that functions just like Windows Explorer does. This defines what the vendor calls the "Carbonite drive". After that Carbonite lurks in the background, sending to the Great Computer In The Sky the data you said you wanted backed up, which is everything in the Carbonite drive as you earlier specified it.

This inititial data upload phase can take quite a long time depending on how much data is at issue. In our case we can pump only about 100MB per hour offsite. If you are on DSL you probably would find yourself with similar throughput. So if you have, say, 1GB of data that you care about, this would be ten hours of transmission time. However, Carbonite runs at low priority, or can be configured to, so with all of your other normal Internet activity the ten hours might stretch out to twenty, for example.

That's a day, and a day is not too terribly bad, but if you have 5GB of data we're now talking about a working week, and if you have multiple 20GB data sets as I did, each data set could take two weeks to image, as was happening with me. Now ...

At some point Carbonite will have backed up all of your data initially, everything in the Carbonite drive. After that it monitors your file system activity, sending to offsite storage each and every file that you change.

In other words, the "Carbonite drive" tries to be a mirror of that portion of your hard drive that you told it to deal with. Trouble is, unless they have changed the philosophy since I worked with the first release of the system in February of 2007, when you delete stuff it will also get deleted from the offsite mirror. (I know that today they keep back versions of changed files, but I don't know what happens today if you actually delete a file.)

So my first problem was that Carbonite was not an archiving system, simply a mirror. There is a workaround of sorts, -- build the archive locally on hard drive, let Carbonite mirror it, and don't delete anything from it. However, this requires increasing amounts of local hard drive capacity, exactly one of the things I'm trying to avoid. Perhaps the system works differently today but this certainly was a problem for me back then.

My second problem was more serious. Whether it is no longer true I don't know, but at the time I was trying to use Carbonite it was clear that it was mechanizing the process roughly as follows: First it builds a list of files to be sent off-site. Each time it decides to back up a file it makes an entry in the list, and it was clear to me that at the time the system was doing a linear search. Since I was eventually trying to back up 600,000 files, the search times became outrageous. In fact, I was generating changed data faster than Carbonite could update its list.

And this slow upload process resulted in a third problem for me -- it was impossible to tell what had been backed up and what had not. In other words, what Carbonite was sending out was not a system snapshot but rather a rolling backup of uncertain composition.

Once I realized that all three problems existed I abandoned Carbonite. However, just because the system wasn't adequate for me doesn't mean that it wouldn't have been adequate for other people, and it is possible that today the system might do exactly what I would want it to do. I don't know. But I also don't really care. I have my own backup procedures that allow me to do true snapshots, and I have my own way of getting data up to a different Great Computer In The Sky, and I'm satisfied for now.

Carbonite was then and surely is by now suitable for my wife, who has about 4GB of data that changes only very, very slowly, mostly in the form of her evolving email archive. So she could make effective use of it, and I'm thinking about taking out another Carbonite subscription just for her so I won't have to deal with backing her stuff up as well as mine.

And there you have it, loki. You mentioned a competing service. I know nothing about it, and all that I know about Carbonite is the way the system stood eighteen months ago. But I will assert that it is ideal for people who would not otherwise be doing backup, and who have amounts of slowly changing data that are typical of most office and home PC users.

When I realized that this was happening I abandoned the system