|| ||Ryo Tsuruta <email@example.com>|
|| ||firstname.lastname@example.org, email@example.com,
|| ||[PATCH 0/2] dm-ioband: I/O bandwidth controller v1.1.0:
|| ||Mon, 02 Jun 2008 21:34:33 +0900 (JST)|
This is dm-ioband version 1.1.0 release.
Dm-ioband is an I/O bandwidth controller implemented as a device-mapper
driver, which gives specified bandwidth to each job running on the same
- Can be applied to the kernel 2.6.26-rc2-mm1.
- Changes from 1.0.0 (posted on May 19, 2008):
- Measures against high memory pressure.
I/O requests to reclaim pages won't be blocked in dm-ioband
even when the group doesn't have the right to issue any more I/Os.
But it can keep to control the bandwidth of the group since these
I/Os will be counted as debt of the group.
- Performance tuning
Improve the performance when a lot of ioband groups are created.
It is useful when you want to control bandwidth on a per user
basis. For example, you can create an ioband group and give an
weight to a user once the user is logged in. The combination of
dm-ioband, the cgroup memory resource controller and the
CPU resource controller will work more effectively.
I also announce that we have launched "Linux Block I/O Bandwidth
Control Project" website http://people.valinux.co.jp/~ryov/bwctl/ .
The goal of this project is to develop a block I/O bandwidth
controller on Linux. You can find information about the followings:
This device mapper driver.
An administration tool for dm-ioband, which makes it easy to
configure the dm-ioband device.
- Block I/O Tracking
Block I/O tracking, which we call "bio cgroup", provides block I/O
tracking mechanism. It makes dm-ioband be able to control the block
I/O bandwidths even when it accepts delays write requests.