I was in the process of setting up another new iMac for a user at work the other day, and got to looking at the hard drive — the ‘entry level’ 24″ iMac comes standard with a dual-core 2.66GHz processor, 4GB of memory and a 600GB hard drive. Much of that capacity (other than that memory) just won’t get used. There will be times when the processors will peak a bit, but most of the time they’ll be just barely above idle. And the hard drive… 600GB? On a desktop machine? If the computer were used in a home setting, that might get utilized, but here… Boy could I use some of that capacity for other stuff on the network! I guess I could just buy some cheap 100GB SATA drives and swap them out, but I’ve seen the gymnastics necessary to replace a drive in an iMac, and I don’t want to go through that any more than absolutely necessary.
I remember back when Apple was first rolling out OS X, there was talk of these super apps that would allow us to tap into some of that unused processing power by creating a distributed network computer by linking the computers on a network together; if one computer had a huge task of some sort to complete and other computers on the network had spare processor cycles available, there’d be some sharing going on, and you could get more done. At least that was the idea, but I haven’t heard much about distributed computer grid clusters since the big splash about using a host of Macs to create a monster grid computer. Xgrid sharing lives on, and even has a checkbox to enable it in the Sharing Preference Panel in Mac OS 10.5 (and maybe earlier.) Years ago when it might take a raster image processor (RIP) multiple hours to chew through an eight-page layout I would’ve have loved to put something like this to work, but today with the typical tasks done on the typical desktop computer in a print shop or an office environment, and without some monstrously processor-intensive task that needs to be done, I don’t really see much point in messing with it.
What I would like to see though is some kind of distributed disk sharing; that iMac I set up today starts out with a whopping big 600GB drive; after loading all the software on it there was still an easy 500GB… And that computer is one of three that I set up recently, and one of five of the same configuration. If I were to partition the disks in each of those machines to set aside half of the available space I’d have an easy terabyte and a half of disk space that could be used for other stuff.
What if there was some way of joining the disks on multiple computers over the network to create a disk array of sorts… A local area network RAID array. Think of a RAID array with the network acting as the interface card and some software on a server striping the bits & bytes across the disks. In all my digging through Google and other search engines, I haven’t found anything like what I’m thinking of; either I’m not asking the right questions or it hasn’t been done yet. If not, that’s too bad, because I think there’s a lot of potential there, but I can also understand some of the obstacles to making it work. The biggest issue is probably that the network can be a lot more fragile than the hardware & software that it takes to make a RAID array in a server or external box work. A mirrored drive in a RAID 1 arrangement would probably work best, as the other RAID levels with the data striped across multiple volumes would require a higher level of availability for the disks than might be possible.
But you know, since it doesn’t look like using that disk space for live files will work any time soon, maybe I can still put it to use for backups; set Retrospect up to use that space for backing stuff on the server up to disk, just for extra redundancy… Hmmm… Might have to play with that a bit…