Based on kernel version 4.9. Page generated on 2016-12-21 14:28 EST.
1 CFQ (Complete Fairness Queueing) 2 =============================== 3 4 The main aim of CFQ scheduler is to provide a fair allocation of the disk 5 I/O bandwidth for all the processes which requests an I/O operation. 6 7 CFQ maintains the per process queue for the processes which request I/O 8 operation(synchronous requests). In case of asynchronous requests, all the 9 requests from all the processes are batched together according to their 10 process's I/O priority. 11 12 CFQ ioscheduler tunables 13 ======================== 14 15 slice_idle 16 ---------- 17 This specifies how long CFQ should idle for next request on certain cfq queues 18 (for sequential workloads) and service trees (for random workloads) before 19 queue is expired and CFQ selects next queue to dispatch from. 20 21 By default slice_idle is a non-zero value. That means by default we idle on 22 queues/service trees. This can be very helpful on highly seeky media like 23 single spindle SATA/SAS disks where we can cut down on overall number of 24 seeks and see improved throughput. 25 26 Setting slice_idle to 0 will remove all the idling on queues/service tree 27 level and one should see an overall improved throughput on faster storage 28 devices like multiple SATA/SAS disks in hardware RAID configuration. The down 29 side is that isolation provided from WRITES also goes down and notion of 30 IO priority becomes weaker. 31 32 So depending on storage and workload, it might be useful to set slice_idle=0. 33 In general I think for SATA/SAS disks and software RAID of SATA/SAS disks 34 keeping slice_idle enabled should be useful. For any configurations where 35 there are multiple spindles behind single LUN (Host based hardware RAID 36 controller or for storage arrays), setting slice_idle=0 might end up in better 37 throughput and acceptable latencies. 38 39 back_seek_max 40 ------------- 41 This specifies, given in Kbytes, the maximum "distance" for backward seeking. 42 The distance is the amount of space from the current head location to the 43 sectors that are backward in terms of distance. 44 45 This parameter allows the scheduler to anticipate requests in the "backward" 46 direction and consider them as being the "next" if they are within this 47 distance from the current head location. 48 49 back_seek_penalty 50 ----------------- 51 This parameter is used to compute the cost of backward seeking. If the 52 backward distance of request is just 1/back_seek_penalty from a "front" 53 request, then the seeking cost of two requests is considered equivalent. 54 55 So scheduler will not bias toward one or the other request (otherwise scheduler 56 will bias toward front request). Default value of back_seek_penalty is 2. 57 58 fifo_expire_async 59 ----------------- 60 This parameter is used to set the timeout of asynchronous requests. Default 61 value of this is 248ms. 62 63 fifo_expire_sync 64 ---------------- 65 This parameter is used to set the timeout of synchronous requests. Default 66 value of this is 124ms. In case to favor synchronous requests over asynchronous 67 one, this value should be decreased relative to fifo_expire_async. 68 69 group_idle 70 ----------- 71 This parameter forces idling at the CFQ group level instead of CFQ 72 queue level. This was introduced after a bottleneck was observed 73 in higher end storage due to idle on sequential queue and allow dispatch 74 from a single queue. The idea with this parameter is that it can be run with 75 slice_idle=0 and group_idle=8, so that idling does not happen on individual 76 queues in the group but happens overall on the group and thus still keeps the 77 IO controller working. 78 Not idling on individual queues in the group will dispatch requests from 79 multiple queues in the group at the same time and achieve higher throughput 80 on higher end storage. 81 82 Default value for this parameter is 8ms. 83 84 low_latency 85 ----------- 86 This parameter is used to enable/disable the low latency mode of the CFQ 87 scheduler. If enabled, CFQ tries to recompute the slice time for each process 88 based on the target_latency set for the system. This favors fairness over 89 throughput. Disabling low latency (setting it to 0) ignores target latency, 90 allowing each process in the system to get a full time slice. 91 92 By default low latency mode is enabled. 93 94 target_latency 95 -------------- 96 This parameter is used to calculate the time slice for a process if cfq's 97 latency mode is enabled. It will ensure that sync requests have an estimated 98 latency. But if sequential workload is higher(e.g. sequential read), 99 then to meet the latency constraints, throughput may decrease because of less 100 time for each process to issue I/O request before the cfq queue is switched. 101 102 Though this can be overcome by disabling the latency_mode, it may increase 103 the read latency for some applications. This parameter allows for changing 104 target_latency through the sysfs interface which can provide the balanced 105 throughput and read latency. 106 107 Default value for target_latency is 300ms. 108 109 slice_async 110 ----------- 111 This parameter is same as of slice_sync but for asynchronous queue. The 112 default value is 40ms. 113 114 slice_async_rq 115 -------------- 116 This parameter is used to limit the dispatching of asynchronous request to 117 device request queue in queue's slice time. The maximum number of request that 118 are allowed to be dispatched also depends upon the io priority. Default value 119 for this is 2. 120 121 slice_sync 122 ---------- 123 When a queue is selected for execution, the queues IO requests are only 124 executed for a certain amount of time(time_slice) before switching to another 125 queue. This parameter is used to calculate the time slice of synchronous 126 queue. 127 128 time_slice is computed using the below equation:- 129 time_slice = slice_sync + (slice_sync/5 * (4 - prio)). To increase the 130 time_slice of synchronous queue, increase the value of slice_sync. Default 131 value is 100ms. 132 133 quantum 134 ------- 135 This specifies the number of request dispatched to the device queue. In a 136 queue's time slice, a request will not be dispatched if the number of request 137 in the device exceeds this parameter. This parameter is used for synchronous 138 request. 139 140 In case of storage with several disk, this setting can limit the parallel 141 processing of request. Therefore, increasing the value can improve the 142 performance although this can cause the latency of some I/O to increase due 143 to more number of requests. 144 145 CFQ Group scheduling 146 ==================== 147 148 CFQ supports blkio cgroup and has "blkio." prefixed files in each 149 blkio cgroup directory. It is weight-based and there are four knobs 150 for configuration - weight[_device] and leaf_weight[_device]. 151 Internal cgroup nodes (the ones with children) can also have tasks in 152 them, so the former two configure how much proportion the cgroup as a 153 whole is entitled to at its parent's level while the latter two 154 configure how much proportion the tasks in the cgroup have compared to 155 its direct children. 156 157 Another way to think about it is assuming that each internal node has 158 an implicit leaf child node which hosts all the tasks whose weight is 159 configured by leaf_weight[_device]. Let's assume a blkio hierarchy 160 composed of five cgroups - root, A, B, AA and AB - with the following 161 weights where the names represent the hierarchy. 162 163 weight leaf_weight 164 root : 125 125 165 A : 500 750 166 B : 250 500 167 AA : 500 500 168 AB : 1000 500 169 170 root never has a parent making its weight is meaningless. For backward 171 compatibility, weight is always kept in sync with leaf_weight. B, AA 172 and AB have no child and thus its tasks have no children cgroup to 173 compete with. They always get 100% of what the cgroup won at the 174 parent level. Considering only the weights which matter, the hierarchy 175 looks like the following. 176 177 root 178 / | \ 179 A B leaf 180 500 250 125 181 / | \ 182 AA AB leaf 183 500 1000 750 184 185 If all cgroups have active IOs and competing with each other, disk 186 time will be distributed like the following. 187 188 Distribution below root. The total active weight at this level is 189 A:500 + B:250 + C:125 = 875. 190 191 root-leaf : 125 / 875 =~ 14% 192 A : 500 / 875 =~ 57% 193 B(-leaf) : 250 / 875 =~ 28% 194 195 A has children and further distributes its 57% among the children and 196 the implicit leaf node. The total active weight at this level is 197 AA:500 + AB:1000 + A-leaf:750 = 2250. 198 199 A-leaf : ( 750 / 2250) * A =~ 19% 200 AA(-leaf) : ( 500 / 2250) * A =~ 12% 201 AB(-leaf) : (1000 / 2250) * A =~ 25% 202 203 CFQ IOPS Mode for group scheduling 204 =================================== 205 Basic CFQ design is to provide priority based time slices. Higher priority 206 process gets bigger time slice and lower priority process gets smaller time 207 slice. Measuring time becomes harder if storage is fast and supports NCQ and 208 it would be better to dispatch multiple requests from multiple cfq queues in 209 request queue at a time. In such scenario, it is not possible to measure time 210 consumed by single queue accurately. 211 212 What is possible though is to measure number of requests dispatched from a 213 single queue and also allow dispatch from multiple cfq queue at the same time. 214 This effectively becomes the fairness in terms of IOPS (IO operations per 215 second). 216 217 If one sets slice_idle=0 and if storage supports NCQ, CFQ internally switches 218 to IOPS mode and starts providing fairness in terms of number of requests 219 dispatched. Note that this mode switching takes effect only for group 220 scheduling. For non-cgroup users nothing should change. 221 222 CFQ IO scheduler Idling Theory 223 =============================== 224 Idling on a queue is primarily about waiting for the next request to come 225 on same queue after completion of a request. In this process CFQ will not 226 dispatch requests from other cfq queues even if requests are pending there. 227 228 The rationale behind idling is that it can cut down on number of seeks 229 on rotational media. For example, if a process is doing dependent 230 sequential reads (next read will come on only after completion of previous 231 one), then not dispatching request from other queue should help as we 232 did not move the disk head and kept on dispatching sequential IO from 233 one queue. 234 235 CFQ has following service trees and various queues are put on these trees. 236 237 sync-idle sync-noidle async 238 239 All cfq queues doing synchronous sequential IO go on to sync-idle tree. 240 On this tree we idle on each queue individually. 241 242 All synchronous non-sequential queues go on sync-noidle tree. Also any 243 request which are marked with REQ_NOIDLE go on this service tree. On this 244 tree we do not idle on individual queues instead idle on the whole group 245 of queues or the tree. So if there are 4 queues waiting for IO to dispatch 246 we will idle only once last queue has dispatched the IO and there is 247 no more IO on this service tree. 248 249 All async writes go on async service tree. There is no idling on async 250 queues. 251 252 CFQ has some optimizations for SSDs and if it detects a non-rotational 253 media which can support higher queue depth (multiple requests at in 254 flight at a time), then it cuts down on idling of individual queues and 255 all the queues move to sync-noidle tree and only tree idle remains. This 256 tree idling provides isolation with buffered write queues on async tree. 257 258 FAQ 259 === 260 Q1. Why to idle at all on queues marked with REQ_NOIDLE. 261 262 A1. We only do tree idle (all queues on sync-noidle tree) on queues marked 263 with REQ_NOIDLE. This helps in providing isolation with all the sync-idle 264 queues. Otherwise in presence of many sequential readers, other 265 synchronous IO might not get fair share of disk. 266 267 For example, if there are 10 sequential readers doing IO and they get 268 100ms each. If a REQ_NOIDLE request comes in, it will be scheduled 269 roughly after 1 second. If after completion of REQ_NOIDLE request we 270 do not idle, and after a couple of milli seconds a another REQ_NOIDLE 271 request comes in, again it will be scheduled after 1second. Repeat it 272 and notice how a workload can lose its disk share and suffer due to 273 multiple sequential readers. 274 275 fsync can generate dependent IO where bunch of data is written in the 276 context of fsync, and later some journaling data is written. Journaling 277 data comes in only after fsync has finished its IO (atleast for ext4 278 that seemed to be the case). Now if one decides not to idle on fsync 279 thread due to REQ_NOIDLE, then next journaling write will not get 280 scheduled for another second. A process doing small fsync, will suffer 281 badly in presence of multiple sequential readers. 282 283 Hence doing tree idling on threads using REQ_NOIDLE flag on requests 284 provides isolation from multiple sequential readers and at the same 285 time we do not idle on individual threads. 286 287 Q2. When to specify REQ_NOIDLE 288 A2. I would think whenever one is doing synchronous write and not expecting 289 more writes to be dispatched from same context soon, should be able 290 to specify REQ_NOIDLE on writes and that probably should work well for 291 most of the cases.