Pros:
* Possible to have network boot with failover to disk of network not available
* Network boot provides single point of control
* Switching a boot OS/kernel/image is as easy as changing out a file on the server and restarting
* Network booting enables the option of having no HD in nodes at all, eliminating moving parts and increasing reliability (I think this will be a significant factor in large scale systems in the future, especially when network boot technologies mature more)
* Common sense and logic: why spend time and effort making a hundred disks look identical when you can use a single image on the network?
Cons:
* Having two bootable devices is more to maintain and keep in sync, and could cause confusion in troubleshooting resulting problems (if it's unrealized that the node booted from an unexpected device)
* Definite scaling limitations (there is lots of room for more development in the network boot arena)
* Unnecessary network traffic if you're looking at ways of minimizing traffic on a shared network
* If using binaries on the node, local disk typically provides much better performance than network mounted (e.g. some users my run tasks like compiling on compute nodes)
* A network booted cluster is much more sensitive to a network outage than a disk booted cluster. Stability or fault tolerance, whichever you wish to label it.
That's about all that occurs to me at the moment-
Jeremy
At 11:34 PM 1/9/2003 -0500, Benoit des Ligneris wrote:
Hello,
* Maximilian A. Ott <[EMAIL PROTECTED]> [03-01-09 23:12]:
> Anyhow, can you guys give me some pro and con on why I should/shouldn't
> always boot over the network.
>From my point of view, no network, no cluster. As a consequence, I
always boot from the network. It gives you the maximum flexibility as
you can still boot from hard drive (second boot device so if there is no
ankswer for PXE it will just boot from hard disk). You can
control everything from the master node (ie boot network
(diskless/systemless), reinstall hard disk, boot hard disk) without moving from your chair ;-)))
> My ideal configuration always was a netboot with a local disk for a cache FS
> and a scratch area.
Same thing for me ;-)))
The thin-OSCAR workgroup was created for this purpose. Please visit
http://thin-oscar.ccs.usherbrooke.ca/
Ben
--
Benoit des Ligneris Etudiant au Doctorat -- Ph. D. Student
Web : http://benoit.des.ligneris.net/
Vice-President du GULUS vice-president http://www.gulus.org/
Mydynaweb Developpe(u)r: http://mydynaweb.net/
GPG/PGP Key http://benoit.des.ligneris.net/linux/gpg.txt
-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Oscar-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/oscar-users
------------------------------------------------------- This SF.NET email is sponsored by: SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See! http://www.vasoftware.com _______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
