About a month ago I wrote a post on Mozy, a new backup tool that I had been using for a few weeks at the time. Well, I thought I'd post an update, since as of 5 hours ago, my initial backup set is finally complete.
Perhaps it seems egregious for an initial backup set to take a month and a half, but I was backing up 59.3 GB of data. About ten years worth of my computer art materials, (legal) music, home video clips, numberless digital photos taken by my wife, and of course my writing. The whole process took longer than it otherwise would have because I throttled it down to about 30 Kbps between 8:30 AM and 10:00 PM, so it was only able to run at a full (for my cable connection) ~390 Kbps during the night hours.
I must say, it's been a bit interesting checking in each day to see how much progress had been made. But still, I'm looking forward to having this just kick off at 2AM every night, and quickly back up whatever changes I've made during the day. That's what I got this for, after all.
If I had one main suggestion for Mozy regarding initial backup sets, it would be this: how about creating a service by which clients can email you DVDs for you to upload straight from your data center? The indexes would certainly be a problem since you normally map everything to the hard disks, but if you made it so that Mozy had the option to do an initial backup set to a set of DVDs, then you could embed the indexes right on the DVDs and easily read them when the clients send them in.
This, of course, would take development time away from your main efforts, and would also require ongoing man hours for staff to handle the DVDs as they arrive. So I wouldn't expect this service to be free -- perhaps $5 per DVD would work for all parties. Whatever covers your costs while making it easier/faster for clients to do that initial backup set. Although, I can see where that added service might cause capacity planning problems in your data center, because your volume of data could quickly jump upwards when a lot of DVDs come in, whereas right now every individual client stream arrives at a fairly predictable trickle.
There are always so many different concerns when you have a centrally hosted software service like this. Part of what professionally interests me in that model, I guess.