1 Deadline IO scheduler tunables
2 ==============================
4 This little file attempts to document how the deadline io scheduler works.
5 In particular, it will clarify the meaning of the exposed tunables that may be
6 of interest to power users.
8 Selecting IO schedulers
9 -----------------------
10 Refer to Documentation/block/switching-sched.txt for information on
11 selecting an io scheduler on a per-device basis.
14 ********************************************************************************
20 The goal of the deadline io scheduler is to attempt to guarantee a start
21 service time for a request. As we focus mainly on read latencies, this is
22 tunable. When a read request first enters the io scheduler, it is assigned
23 a deadline that is the current time + the read_expire value in units of
30 Similar to read_expire mentioned above, but for writes.
36 When a read request expires its deadline, we must move some requests from
37 the sorted io scheduler list to the block device dispatch queue. fifo_batch
38 controls how many requests we move.
41 writes_starved (number of dispatches)
44 When we have to move requests from the io scheduler queue to the block
45 device dispatch queue, we always give a preference to reads. However, we
46 don't want to starve writes indefinitely either. So writes_starved controls
47 how many times we give preference to reads over writes. When that has been
48 done writes_starved number of times, we dispatch some writes based on the
49 same criteria as reads.
55 Sometimes it happens that a request enters the io scheduler that is contigious
56 with a request that is already on the queue. Either it fits in the back of that
57 request, or it fits at the front. That is called either a back merge candidate
58 or a front merge candidate. Due to the way files are typically laid out,
59 back merges are much more common than front merges. For some work loads, you
60 may even know that it is a waste of time to spend any time attempting to
61 front merge requests. Setting front_merges to 0 disables this functionality.
62 Front merges may still occur due to the cached last_merge hint, but since
63 that comes at basically 0 cost we leave that on. We simply disable the
64 rbtree front sector lookup when the io scheduler merge function is called.
67 Nov 11 2002, Jens Axboe <jens.axboe@oracle.com>