Thanks for the plug for my old USENIX paper :-). Tomorrow I lecture my architecture students on I/O systems, and somehow I doubt I’ll get past talking about disk drives in an hour and a half — they’re just too fascinating. I owe it to them to talk about buses, device drivers, RAID, and some other things, too, but that may have to wait…

I ran across your ZCAV notes a while back, too. Your work is very thorough, I’m really impressed.

I think to see a real deviation from linear in the performance, you have to have a high outer:inner diameter ratio, but I would expect that to be low in laptop drives. Hmm.

It’s true that you can see the zoning effects even through the file system, which was the point of my paper, but I started with a large, mostly empty partition. You can see weird effects in SunOS’s allocation policy as the cylinder groups start to fill.

I think I still have some of the source code for those tools around somewhere. Keith Smith and John DiMarco were kind enough to share, then I modified them somewhat to get what I needed.

–Rod