> Sure, but then that defeats the whole point of a "$300 pee cee".  At 
> that point I might as well spend $4000 on a power server as I'm likely 
> to be spending $2000 (or so) on the pee cee and $700 on the card and 
> still only getting a Band-Aid solution.

Let's be specific.  What exactly would you spend that $4000 on?  What is
a "power server"?

If you think the IBM solution is Band-Aid I suggest you have a second
look at the technology.  IPMI _is_ band-aid.  Remember that IBM still is
a mainframe company.  Andrew, what specifically are you missing that 
reduces this to "Band-Aid"?

This thing is a little embedded CPU on the card that does out-of-band
management, and does it well.

Look at Chapter 3: Remote Supervisor Adapter II of this book:
http://www.redbooks.ibm.com/abstracts/sg246495.html?Open
In short:
*  Automatic notification and alerts        <-- it'll page you
*  Continuous health monitoring and control <-- temperature
*  Event log                                <-- logs in nvram
*  Operating system failure screen capture
*  Remote media                             <-- put a cd or floppy into your 
desktop and it'll mount remotely
*  Remote power control                     <-- reboot

The guys I know that run a Top-500 linux clusters swear by these RSA 
cards, used along with xCAT -- an open system management platform 
for maintaining clusters:

http://www.alphaworks.ibm.com/tech/xCAT/

The corporate side that don't need to dig into the details go with
IBM Director to manage them instead of xCAT.

> Side note: does anybody know how the Apple servers handle remote 
> management and consoles?  It's an option that I just don't know anything 
> about.

With Apple servers you're stuck with either the "Server Administrator", 
VNC, and ssh only.  Out-of-band management doesn't exist, if it
crashes, you're no better off than you'd be with a $300 pee-cee -- you
drive at 100 mph to the colo as you boss's boss tells customers it'll
be back up soon.  When you get there you hope and pray that something
made it into the syslog that will tell you what happened.



-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to