On Tue, Jul 09, 2013 at 06:45:58AM -0700, Tejun Heo wrote: > On Tue, Jul 09, 2013 at 05:43:10PM +0400, Konstantin Khlebnikov wrote: > > My concept it cgroup which would control io operation on vfs layer > > for all filesystems. It will account and manage IO operations. I've > > found really lightweight technique for accounting and throttling > > which don't introduce new locks or priority inversions (which is > > major problem in all existing throttlers, including cpu cgroup rate > > limiter) So, I've tried to keep code smaller, cleaner and saner as > > possible while you guys are trying to push me into the block layer > > =) > > You're trying to implement QoS in the place where you don't have > control of the queue itself. > You aren't even managing the right type > of resource for disks which is time slice rather than iops or > bandwidth For implementing throttling one as such does not have to do time slice management on the queue. For providing constructs like IOPS or bandwidth throttling, one just need to put one throttling knob in the cgroup pipe irrespective of time slice management on the backing device/network. Also time slice management is one way of managing the backend resource. CFQ did that and it works only for slow devices. For faster devices we anyway need some kind of token mechanism instead of keeping track of time. So I don't think trying to manage time slice is the requirement here. > and by the time you implemented proper hierarchy support and > proportional control, yours isn't gonna be that simple either. I suspect he is not plannnig to do any proportional control at that layer. Just throttling mechanism. Thanks Vivek -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>