diff options
Diffstat (limited to 'Documentation/block')
-rw-r--r-- | Documentation/block/deadline-iosched.txt | 14 |
1 files changed, 10 insertions, 4 deletions
diff --git a/Documentation/block/deadline-iosched.txt b/Documentation/block/deadline-iosched.txt index c23cab13c3d1..72576769e0f4 100644 --- a/Documentation/block/deadline-iosched.txt +++ b/Documentation/block/deadline-iosched.txt | |||
@@ -30,12 +30,18 @@ write_expire (in ms) | |||
30 | Similar to read_expire mentioned above, but for writes. | 30 | Similar to read_expire mentioned above, but for writes. |
31 | 31 | ||
32 | 32 | ||
33 | fifo_batch | 33 | fifo_batch (number of requests) |
34 | ---------- | 34 | ---------- |
35 | 35 | ||
36 | When a read request expires its deadline, we must move some requests from | 36 | Requests are grouped into ``batches'' of a particular data direction (read or |
37 | the sorted io scheduler list to the block device dispatch queue. fifo_batch | 37 | write) which are serviced in increasing sector order. To limit extra seeking, |
38 | controls how many requests we move. | 38 | deadline expiries are only checked between batches. fifo_batch controls the |
39 | maximum number of requests per batch. | ||
40 | |||
41 | This parameter tunes the balance between per-request latency and aggregate | ||
42 | throughput. When low latency is the primary concern, smaller is better (where | ||
43 | a value of 1 yields first-come first-served behaviour). Increasing fifo_batch | ||
44 | generally improves throughput, at the cost of latency variation. | ||
39 | 45 | ||
40 | 46 | ||
41 | writes_starved (number of dispatches) | 47 | writes_starved (number of dispatches) |