I heard about it and i wanted to know is it worth or not.
It really helped.
Regards
On Tue, Dec 16, 2008 at 12:26 AM, Chrissie Caulfield <ccaulfie@xxxxxxxxxx> wrote:
I don't know about that particular project but I do know something aboutAli Nazemian wrote:
> Hi every body,
> Any body knows that what happened to C-Sharifi project that announce 1
> year ago?! that was about implementing clustering management in kernel...
> Cheers.
implementing clustering in the kernel (from RHEL4 cman) and generally
it's a waste of time and energy. The cluster manager is not
performance-critical enough and too complex to do sensibly in the kernel.
Having all the infrastructure in the kernel simply makes developing and
testing much much harder for no real gain. The performance-critical
components (eg lock manager and filesystem) can still be in the kernel
and leave the cluster manager in userspace where it can be easily
managed and reconfigured - have you ever tried to get an XML parser into
the kernel? No, me neither and I don't want to try ;-)
About the only thing that a kernel-based cluster manager gains you is
freedom from memory inversion problems in low memory situations as you
can guarantee it is always in physical RAM, but even that isn't total as
you still have to rely on the networking layers. And there are ways of
locking userland process into memory anyway. They're not quite as
effective as having it kernel-based but those aren't perfect either!
Just my ha-penny's worth ...
Chrissie
--
Linux-cluster mailing list
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
--
A.Nazemian
-- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
