On 08/05/2013 06:52 PM, Kevin Abbey wrote:
I started using cgroups for control memory usage last week. One user
reported his application takes 4 times longer to complete. I read
elsewhere that cgroup mem. control can reduce performance. Is this
amount realistic?
Is there a more efficient method to control memory usage on nodes which
are shared?
Hi Kevin,
We're running Scientific Linux 6 and ran into some similar performance
issues. When upgrading from the SL 6.2 to 6.3 kernel (2.6.32-220 to
2.6.32-358), we saw significant performance improvements with the cgroup
memory controller. I think the last time we did testing, overhead went
from 25% (or worse, in some cases) down to less than 5%.
We found a couple of kernel commits that really made a difference:
https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=569b846d
https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=cdec2e42
Happily, they are both included in the 2.6.32-358 and later kernels.
Regards,
John
