10:00 AM Geez. I think I'm finally done with being sick. Went home on Thursday and discovered I had, once again, a fever of 102. Joy. So I had plenty of time this weekend to do ABSOLUTELY NOTHING.

I did manage to get some reading done - Inside Windows 2000 Server, by William Boswell. So new the ink was still wet when I picked it up. From what I've read so far, it's an excellent book, very well written. Everything that I've picked up on about Windows 2000 is in there, and a lot I didn't know. If you're going to be running Windows 2000 Server at all, you need this book. Highly recommended.

And speaking of which, I'm off to play with Exchange. Later all.

1:00 PM Wow! Two whole days of conciousness! I'm on a roll!!

I'm also busy. Exchange server games, two new hardware servers to install and configure, one old machine to be replaced - WITHOUT losing any of the extremely critical data it contains. On top of which, I need to maintain iTOOL's servers and keep them all running, and my fellow SysAdmin is in Austin, Texas for a customer installation. Goody.

Other than that, not much is going on. The new apartment is wonderful, particularly for sleeping. I know this because I just spent the last week asleep. <G>

And, finally, an offline discussion prompted some ponderings on flaming. How useful is it? Does it have any use at all? I had never thought so, but this person I was talking to made an excellent point - it does tend to make people think about what they say before they say it, online at least. Think about it. How quickly are you going to spout off an opinion if you know that it'll be flamed if it isn't well thought out? Discuss...

This day devoured by rabid llamas. Twice.

I wrote an update here. I did. A good one. Full of lots of info on Benchmarking Windows 2000 on various kinds of hardware. I wrote, saved it, and copied it to the server. Those wacko Linux nuts must have deleted the update. OR my HTML editor hosed it for me. That's also possible, I suppose. But less fun to think about.

8:20 AMSee above. I did indeed post something yesterday, but it vanished to wherever socks and lost data go. Strange.

Anyway, I mentioned that I was benchmarking Windows 2000 on two different machines. I got different results from Bob, and we've been discussing them by email. Here're my results, compared with NT 4:

-----Original Message-----
From: Matt Beland [mailto:matt@rearviewmirror.org]
Sent: Friday, January 21, 2000 7:15 AM To: Gang@Daynotes. Com Subject: Win2k Gold Release

Bob,

>>>
I know this flies in the face of conventional wisdom, but W2KP seems noticeably slower than NT4. That's not supposed to be. W2KP definitely seems slower on kiwi, but that doesn't surprise me. Beta releases are filled with debugging code that makes them slow. I did expect the release version to be faster than NT4, though.
<<<

The numbers I've gotten the last few days are the opposite; Win2k is FASTER than NT4, on a given piece of hardware. I think that will flip-flop on low-end systems, particularly those with the bare-minimum Win2k requirement of 64 MB of RAM. Now granted, that's only true with 2195 - RC2 and RC3 are both slower than NT, but as you say, that's most likely because of the debugging code. I wonder where the difference is that's making your system slower?

BTW, my "benchmarks" are simple timed operations; open a document in Word from a "cold start", open a document with Word already open, render a "standard" image in Photoshop, print a document, and copy a file that is exactly 100 MB from CD to Disk.

-----Original Message-----
From: Robert Bruce Thompson [mailto:thompson@ttgnet.com] Sent: Friday, January 21, 2000 7:29 AM
To: 'Matt Beland'
Cc: gang@daynotes.com; jerryp@jerrypournelle.com
Subject: RE: Win2k Gold Release

Well, I haven't run any benchmarks. What I reported was simply my impressions of speed. But I think they're valid nonetheless. I work in Word, Outlook, and FrontPage all day long, and W2KP seems slower on the three systems I've run it on (not counting kiwi, which is still running RC2). Those systems are (a) a Pentium III/550 with 128 MB and a Seagate Barracuda U2W SCSI, (b) a Pentium II/300 with 128 MB and an old 4.3 GB IDE, and (c) a Pentium III/450 with 128 MB and a Maxtor 91000D8 10 GB UDMA. All are running NT4W with SP5 or SP6a installed and W2KP 2195.

The difference is noticeable on all of them, but particularly so on the Pentium II/300. Applications take longer to load, and aren't as "snappy" under W2KP. Large data files seem to take longer to load over the network. Web pages take longer to load, and IE itself lags briefly when I click on a link. I don't think this is anything attributable to my hardware or configuration. I think W2KP is simply slower.

But I may be wrong. All of this is purely subjective. It might be interesting to run some real benchmarks.

-----Original Message-----
From: Matt Beland [mailto:matt@rearviewmirror.org] Sent: Friday, January 21, 2000 7:50 AM
To: Robert Bruce Thompson
Cc: gang@daynotes.com; jerryp@jerrypournelle.com
Subject: RE: Win2k Gold Release

As I mentioned on my page yesterday, the systems I ran the benchmarks on are

&sp;&sp;&sp;A) a PII 400 with 256 and a 10 GB IDE on an Intel 440BX-2

&sp;&sp;&sp;B) a dual PII 400 with 256 and 2 IDE drives, a 13 and a 7.x on a TYAN Tiger 100 (440 BX chipset)

The numbers weren't anything spectacular; Win2K Pro was about 3% faster than NT on the single processor machine, and about 5% faster on the dual. In both cases, I compared Win2K Pro with Windows NT Workstation with Service Pack 5 - I don't like 6a. It breaks ColdFusion, and that's our bread-and-butter program around here.

The SERVER comparisons weren't even close. Win2k blew NT out of the water. I think that had a lot to do with the RAID optimizations, though. With NT, just installing on a RAID system (as I'm sure you know) is a pain; some admins I know are happy to get the array configured and the OS loaded, screw optimization. Win2k recognized our standard DELL Perc/2 controller on boot, loaded the arrays, and made snide comments. Well, not that bad, but file transfers using the RAID were as much as 30% faster than NT. We're running a series of tests on IIS5 today; the preliminary results look... interesting, shall we say.