linux/Documentation/block/cfq-iosched.txt

CFQ ioscheduler tunables
========================

slice_idle
----------
This specifies how long CFQ should idle for next request on certain cfq queues
(for sequential workloads) and service trees (for random workloads) before
queue is expired and CFQ selects next queue to dispatch from.

By default slice_idle is a non-zero value. That means by default we idle on
queues/service trees. This can be very helpful on highly seeky media like
single spindle SATA/SAS disks where we can cut down on overall number of
seeks and see improved throughput.

Setting slice_idle to 0 will remove all the idling on queues/service tree
level and one should see an overall improved throughput on faster storage
devices like multiple SATA/SAS disks in hardware RAID configuration. The down
side is that isolation provided from WRITES also goes down and notion of
IO priority becomes weaker.

So depending on storage and workload, it might be useful to set slice_idle=0.
In general I think for SATA/SAS disks and software RAID of SATA/SAS disks
keeping slice_idle enabled should be useful. For any configurations where
there are multiple spindles behind single LUN (Host based hardware RAID
controller or for storage arrays), setting slice_idle=0 might end up in better
throughput and acceptable latencies.

CFQ IOPS Mode for group scheduling
===================================
Basic CFQ design is to provide priority based time slices. Higher priority
process gets bigger time slice and lower priority process gets smaller time
slice. Measuring time becomes harder if storage is fast and supports NCQ and
it would be better to dispatch multiple requests from multiple cfq queues in
request queue at a time. In such scenario, it is not possible to measure time
consumed by single queue accurately.

What is possible though is to measure number of requests dispatched from a
single queue and also allow dispatch from multiple cfq queue at the same time.
This effectively becomes the fairness in terms of IOPS (IO operations per
second).

If one sets slice_idle=0 and if storage supports NCQ, CFQ internally switches
to IOPS mode and starts providing fairness in terms of number of requests
dispatched. Note that this mode switching takes effect only for group
scheduling. For non-cgroup users nothing should change.

CFQ IO scheduler Idling Theory
===============================
Idling on a queue is primarily about waiting for the next request to come
on same queue after completion of a request. In this process CFQ will not
dispatch requests from other cfq queues even if requests are pending there.

The rationale behind idling is that it can cut down on number of seeks
on rotational media. For example, if a process is doing dependent
sequential reads (next read will come on only after completion of previous
one), then not dispatching request from other queue should help as we
did not move the disk head and kept on dispatching sequential IO from
one queue.

CFQ has following service trees and various queues are put on these trees.

	sync-idle	sync-noidle	async

All cfq queues doing synchronous sequential IO go on to sync-idle tree.
On this tree we idle on each queue individually.

All synchronous non-sequential queues go on sync-noidle tree. Also any
request which are marked with REQ_NOIDLE go on this service tree. On this
tree we do not idle on individual queues instead idle on the whole group
of queues or the tree. So if there are 4 queues waiting for IO to dispatch
we will idle only once last queue has dispatched the IO and there is
no more IO on this service tree.

All async writes go on async service tree. There is no idling on async
queues.

CFQ has some optimizations for SSDs and if it detects a non-rotational
media which can support higher queue depth (multiple requests at in
flight at a time), then it cuts down on idling of individual queues and
all the queues move to sync-noidle tree and only tree idle remains. This
tree idling provides isolation with buffered write queues on async tree.

FAQ
===
Q1. Why to idle at all on queues marked with REQ_NOIDLE.

A1. We only do tree idle (all queues on sync-noidle tree) on queues marked
    with REQ_NOIDLE. This helps in providing isolation with all the sync-idle
    queues. Otherwise in presence of many sequential readers, other
    synchronous IO might not get fair share of disk.

    For example, if there are 10 sequential readers doing IO and they get
    100ms each. If a REQ_NOIDLE request comes in, it will be scheduled
    roughly after 1 second. If after completion of REQ_NOIDLE request we
    do not idle, and after a couple of milli seconds a another REQ_NOIDLE
    request comes in, again it will be scheduled after 1second. Repeat it
    and notice how a workload can lose its disk share and suffer due to
    multiple sequential readers.

    fsync can generate dependent IO where bunch of data is written in the
    context of fsync, and later some journaling data is written. Journaling
    data comes in only after fsync has finished its IO (atleast for ext4
    that seemed to be the case). Now if one decides not to idle on fsync
    thread due to REQ_NOIDLE, then next journaling write will not get
    scheduled for another second. A process doing small fsync, will suffer
    badly in presence of multiple sequential readers.

    Hence doing tree idling on threads using REQ_NOIDLE flag on requests
    provides isolation from multiple sequential readers and at the same
    time we do not idle on individual threads.

Q2. When to specify REQ_NOIDLE
A2. I would think whenever one is doing synchronous write and not expecting
    more writes to be dispatched from same context soon, should be able
    to specify REQ_NOIDLE on writes and that probably should work well for
    most of the cases.