Alex Hopmann’s Blog

Several posts from Mac people have complained about the Vista virtualization license rules that basically say you can’t do it unless you have the Ultimate or Business SKUs. Now, I think this is a lame business strategy on Microsoft’s part, but is it unfair as has been alleged?

If its really unfair, how about Apple lets us run MacOS virtualized inside Windows. I challenge Apple to say that they are fine with this. I’d even be fine if they only went half-way and only allowed it if there underlying hardware was a Mac (but I’m still going to call that a 50% cop-out compared to the “Apples to Apples” scenario of buying a retail copy of MacOS and installing it on a VM on a arbitrary PC.

I’m pretty conflicted on what to do for hard-drives for my new workstation this summer.

One thought is to use the Raptor for the system drive. The Raptors are relatively small expensive drives (about $240 for 150GB) that run at 10000 RPM (instead of 7200 or 5600) with fast controllers. It should result in faster boots, application launches, and general performance. Of course I want to balance that against system reliability. From one point of view there is no more painful experience than upgrading your boot drive and/or the drive that contains your “Program Files”. Getting everything back into shape could take days and considering what my time is worth that is certainly worth a few extra hundred dollars.

Tom’s Hardware wrote a comparison of the Raptor vs. RAID arrays for workstation performance. This article is useful for some data but the analysis seems really poor to me and it appears to me from the title (”Cheap RAID ravages WD Raptor”) that it was written by someone trying to prove a pre-conceived point. First of all they use striping-RAID which I wouldn’t touch anymore- the greatly decreased reliability is just too much of a problem since a failure one either drive fails the whole thing. Still, contrary to the “ravages” claim, a single Raptor still beat the RAID array of slower drives by about 50% on key synthetic tests and two Raptors demolished them.

Still the performance advantage only translates to a 21.3 second boot time vs. 23.4 seconds or about a 10% difference in the real-world performance. This is nice enough to be worth considering despite the extra size, but another factor comes in- the 150GB is just too small.

My current workstation machine is using 235GB of 280GB on its boot drive. Program Files is 60 GB and the Windows directory is 8 GB. So this should fit in theory, but of course I’ve got a lot of crap on there that I would probably put somewhere else given better planning. Windows applications are just used to putting their stuff in the normal places on C and every time you try to point them somewhere else, you suffer some strange issues.

I’ll also point out that this disk is not the initial disk I got with this system. I bought an upgrade drive which came with some software that let you transfer your boot disk to the new drive. The transfer worked. Sort-of. But things have never been quite the same. Icons are messed up, program associations are messed up, and its something I’d like to avoid ever doing in the future. This makes getting a huge boot drive seem pretty attractive- even if I sacrifice 10% off various performance, never having to worry about having enough space for all that stuff sounds like a good trade.

I’m also tempted to set the boot drive (whichever one I pick) up as a mirrored array so I don’t have any downtime from the system drive going down. In theory a mirrored array gives you better read performance (important for boot) since sectors can be read from either drive (but unlike a striped array you don’t gain any write performance since each sector needs to be written to both drives).

With all these factors my current thought is to go with two 500GB drives in a mirror array for the boot and one bigger drive (1TB or a 750GB I have sitting around) for the data disk. 500GB should be plenty of room for Program Files, Windows and tons of temporary files, and the price per GB is best in that size range ($.25/GB for 400GB, $.30/GB for 500GB, $.40/GB for 1TB).

The irony is that in the end I wind up with a similar recommendation to that Tom’s article, although with different reasoning and specifics….

When I first placed my order for my Canon TX1 on Amazon they listed the release date as March 31st. About a week later that changed to May 31st and the wait is really gnawing at me. Dell lists the camera as shipping in 1-2 weeks but I’ve seen them be totally full of it on those estimates before. There are some online forums where some people suggested that it would start shipping next week, but I guess I’ll just have to wait and see.

I ordered 2 spare batteries for $20 each from some online cheap battery place and a slow 8GB memory card for $60. I figure with 8GB of memory I’ll usually be running out of batteries long before I run out of storage space. The I calculated that I probably need memory that runs at about 4MB/second to shoot continuous video so I’ll probably have to get another faster memory card later, but they aren’t out yet (and look like they will be 4x the price of the existing memory cards). I’m also a little skeptical about whether the cheap batteries are any good- its one of those things where they cost 1/3 the price of the official ones, and they might be crap or they might be 100% the same item with just different printing on the label. I’ll post back here once I found out how it all works.