Thank you for that. It's a pleasure to find a fellow critical thinker. The academic world is slowly moving away from Derrida's mysticism, but I am continually discouraged by the lack of critical thought among faculty members. For example, the reaction to bigoted speech on campus is to engage in anti free speech measures. Public computers are locked down, students are encouraged to report on one another, and dissent is not tolerated. Rational thought is the first thing to go. I remain hopeful, though, that we are on the verge of a return to reason in the academy.
There are a couple of things that you should be aware of before putting VNC on Windows 2000 servers. The first is the well-known problems with security, including the "encrypted" hash stored in the registry and the brute force vulnerability.
More importantly, though, for those of you thinking of running VNC on heavily-used Windows servers is that the CPU run queue tends to increase by 8 or so when VNC is in full screen polling mode. If you want to run perfmon over a remote connection, the two free RDP connections on W2K Server are a much better choice.
VNC is really terrible connecting to Windows 2000 if you poll the whole screen. On a relatively high-end dual cpu machine I saw the processor queue length jump from three (during a heavy DB2 load) to hover around ten, and the context switches/sec quadruple. If the full screen polling was off, the machine wasn't affected quite as much, but as soon as the mouse was moved, a spike occured. I ran a time test of a load with performance monitor on and VNC poll full screen selected, and it took more than twice the time that a regular load usually does.
The windows 2000 remote control mode didn't affect the machine at all.
Most remote control access that I need to do on windows machines requires me to turn on the full screen polling. I would *not* recommend using VNC on a heavily hit production server.
Actually, on a 40 MIP machine, it feels like a 486/33. The best I could get out of a 100 mbps ethernet adapter on the machine was about 400Kbps, and the 390's cpu was at 100%. I realize I'm using a slow 390, but I don't see how anyone can scale to thousands of machines with acceptable performance.
Slashdot Mirror
This is a bad idea. Biometrics are passwords that can't be changed.
Thank you for that. It's a pleasure to find a fellow critical thinker. The academic world is slowly moving away from Derrida's mysticism, but I am continually discouraged by the lack of critical thought among faculty members. For example, the reaction to bigoted speech on campus is to engage in anti free speech measures. Public computers are locked down, students are encouraged to report on one another, and dissent is not tolerated. Rational thought is the first thing to go. I remain hopeful, though, that we are on the verge of a return to reason in the academy.
That book sounds interesting. Who's the author, and what is the publication date?
Is that an official IBM stance?
There are a couple of things that you should be aware of before putting VNC on Windows 2000 servers. The first is the well-known problems with security, including the "encrypted" hash stored in the registry and the brute force vulnerability.
More importantly, though, for those of you thinking of running VNC on heavily-used Windows servers is that the CPU run queue tends to increase by 8 or so when VNC is in full screen polling mode. If you want to run perfmon over a remote connection, the two free RDP connections on W2K Server are a much better choice.
VNC is really terrible connecting to Windows 2000 if you poll the whole screen. On a relatively high-end dual cpu machine I saw the processor queue length jump from three (during a heavy DB2 load) to hover around ten, and the context switches/sec quadruple. If the full screen polling was off, the machine wasn't affected quite as much, but as soon as the mouse was moved, a spike occured. I ran a time test of a load with performance monitor on and VNC poll full screen selected, and it took more than twice the time that a regular load usually does.
The windows 2000 remote control mode didn't affect the machine at all.
Most remote control access that I need to do on windows machines requires me to turn on the full screen polling. I would *not* recommend using VNC on a heavily hit production server.
Actually, on a 40 MIP machine, it feels like a 486/33. The best I could get out of a 100 mbps ethernet adapter on the machine was about 400Kbps, and the 390's cpu was at 100%. I realize I'm using a slow 390, but I don't see how anyone can scale to thousands of machines with acceptable performance.