continuous leaky bucket throttling
v3 cover continuous leaky bucket throttling
This patchset implement continous leaky bucket throttling.
It works mostly on the general case.
The exception is where the load is composed of both reads and writes and two limits iops_rd and iops_wr are set.
The resulting iops are a little above half of the given limits. I tried various strategies to avoid this: two timer, two throttled request queues or even a different algorithm using a priority queue.
The problem is still the same in every version of the code: reads and writes operation seems entangled.
Benoît Canet (2):
throttle: Add a new throttling API implementing continuus leaky
bucket.
block: Enable the new throttling code in the block layer.
block.c | 316 ++++++++------------------------ block/qapi.c | 21 +-- blockdev.c | 115 ++++++------ include/block/block.h | 1 - include/block/block_int.h | 33 +--- include/qemu/throttle.h | 111 ++++++++++++ util/Makefile.objs | 1 + util/throttle.c | 436 +++++++++++++++++++++++++++++++++++++++++++++ 8 files changed, 698 insertions(+), 336 deletions(-) create mode 100644 include/qemu/throttle.h create mode 100644 util/throttle.c
Signature · 1 line
1.7.10.4
RFC v3 1/2 throttle: Add a new throttling API implementing continuus leaky bucket.
Implement the continuous leaky bucket algorithm devised on IRC as a separate module.
--- include/qemu/throttle.h | 111 ++++++++++++ util/Makefile.objs | 1 + util/throttle.c | 436 +++++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 548 insertions(+) create mode 100644 include/qemu/throttle.h create mode 100644 util/throttle.c
include/qemu/throttle.h new file +111 -0
@@ -0,0 +1,111 @@ +/* +* QEMU throttling infrastructure +* +* Copyright (C) Nodalink, SARL. 2013 +* +* Author: +* Benoît Canet <[email protected]> +* +* This program is free software; you can redistribute it and/or +* modify it under the terms of the GNU General Public License as +* published by the Free Software Foundation; either version 2 or +* (at your option) version 3 of the License. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, see <http://www.gnu.org/licenses/>. +*/ + +#ifndef THROTTLING_H +#define THROTTLING_H + +#include <stdint.h> +#include "qemu-common.h" +#include "qemu/timer.h" + +#define NANOSECONDS_PER_SECOND 1000000000.0 + +#define BUCKETS_COUNT 6 + +typedef enum { +THROTTLE_BPS_TOTAL =0 , +THROTTLE_BPS_READ =1 , +THROTTLE_BPS_WRITE =2 , +THROTTLE_OPS_TOTAL =3 , +THROTTLE_OPS_READ =4 , +THROTTLE_OPS_WRITE =5 , +}BucketType ; + +typedef struct LeakyBucket { +int64_t ups;/* units per second */ +int64_t max;/* leaky bucket max in units */ +double bucket;/* bucket in units */ +int64_t previous_leak;/* timestamp of the last leak done */ +}LeakyBucket ; + +/* The following structure is used to configure a ThrottleState +* It contains a bit of state: the bucket field of the LeakyBucket structure. +* However it allows to keep the code clean and the bucket field is reset to +* zero at the right time. +*/ +typedef struct ThrottleConfig { +LeakyBucket buckets[6 ];/* leaky buckets */ +int64_t unit_size;/* size of an unit in bytes */ +int64_t op_size;/* size of an operation in units */ +}ThrottleConfig ; + +typedef struct ThrottleState { +ThrottleConfig cfg; +bool timer_is_throttling_write;/* is the timer throttling a write */ +QEMUTimer *timer;/* timer used to do the throttling */ +QEMUClock *clock;/* the clock used */ +}ThrottleState ; + +/* following 3 function exposed for tests */ +bool throttle_do_start (ThrottleState *ts, +bool is_write, +int64_t size, +int64_t now, +int64_t *next_timer); + +bool throttle_do_end (ThrottleState *ts, +bool is_write, +int64_t now, +int64_t *next_timer); + +bool throttle_do_timer (ThrottleState *ts, +bool is_write, +int64_t now, +int64_t *next_timer); + +/* user API functions */ +void throttle_init (ThrottleState *ts, +QEMUClock *clock, +void (timer )(void *), +void *timer_opaque); + +void throttle_destroy (ThrottleState *ts); + +bool throttle_start (ThrottleState *ts,bool is_write,int64_t size); + +void throttle_end (ThrottleState *ts,bool is_write); + +void throttle_timer (ThrottleState *ts,int64_t now,bool *must_wait); + +void throttle_config (ThrottleState *ts,ThrottleConfig *cfg); + +void throttle_get_config (ThrottleState *ts,ThrottleConfig *cfg); + +bool throttle_enabled (ThrottleConfig *cfg); + +bool throttle_conflicting (ThrottleConfig *cfg); + +bool throttle_is_valid (ThrottleConfig *cfg); + +bool throttle_have_timer (ThrottleState *ts); + +#endif
util/Makefile.objs +1 -0
@@ -11,3 +11,4 @@ util-obj-y += iov.o aes.o qemu-config.o qemu-sockets.o uri.o notify.o util-obj-y += qemu-option.o qemu-progress.o util-obj-y += hexdump.o util-obj-y += crc32c.o +util-obj-y += throttle.o
util/throttle.c new file +436 -0
436 lines not on this page · raw message
Signature · 1 line
1.7.10.4
RFC v3 2/2 block: Enable the new throttling code in the block layer.
In reply to Benoît Canet--- block.c | 316 +++++++++++---------------------------------- block/qapi.c | 21 ++- blockdev.c | 115 +++++++++-------- include/block/block.h | 1 - include/block/block_int.h | 33 +---- 5 files changed, 150 insertions(+), 336 deletions(-)
block.c +72 -244
@@ -86,13 +86,6 @@ static void coroutine_fn bdrv_co_do_rw(void *opaque); static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,int64_t sector_num,int nb_sectors); -static bool bdrv_exceed_bps_limits (BlockDriverState *bs,int nb_sectors, -bool is_write,double elapsed_time,uint64_t *wait); -static bool bdrv_exceed_iops_limits (BlockDriverState *bs,bool is_write, -double elapsed_time,uint64_t *wait); -static bool bdrv_exceed_io_limits (BlockDriverState *bs,int nb_sectors, -bool is_write,int64_t *wait); - static QTAILQ_HEAD(, BlockDriverState) bdrv_states =QTAILQ_HEAD_INITIALIZER (bdrv_states);
@@ -123,55 +116,68 @@ int is_windows_drive(const char *filename)#endif /* throttling disk I/O limits */ -void bdrv_io_limits_disable (BlockDriverState *bs) +void bdrv_set_io_limits (BlockDriverState *bs, +ThrottleConfig *cfg) { - bs->io_limits_enabled = false; +throttle_config (&bs->throttle_state, cfg); +} -while (qemu_co_enter_next (&bs->throttled_reqs)) { - } +static bool bdrv_drain_throttled (BlockDriverState *bs) +{ +bool drained = false; +int i; -if (bs->block_timer) { -qemu_del_timer (bs->block_timer); -qemu_free_timer (bs->block_timer); - bs->block_timer =NULL ; +for (i =0 ; i <2 ; i++) { + while (qemu_co_enter_next(&bs->throttled_reqs[i]) { + drained = true; + } } - bs->slice_start =0 ; - bs->slice_end =0 ; +return drained; } -static void bdrv_block_timer (void *opaque) +void bdrv_io_limits_disable (BlockDriverState *bs) { -BlockDriverState *bs = opaque; + bs->io_limits_enabled = false; + +bdrv_drain_throttled (bs); -qemu_co_enter_next (&bs->throttled_reqs); +throttle_destroy (&bs->throttle_state); } -void bdrv_io_limits_enable (BlockDriverState *bs) +static void bdrv_throttle_timer_cb (void *opaque) { -qemu_co_queue_init (&bs->throttled_reqs); - bs->block_timer =qemu_new_timer_ns (vm_clock, bdrv_block_timer, bs); - bs->io_limits_enabled = true; +BlockDriverState *bs = opaque; +int now =qemu_get_clock_ns (vm_clock); +bool must_wait[2 ];/* does throttled reads or writes must wait */ +int i; + +throttle_timer (&bs->throttle_state, now, must_wait); + +/* execute not throttled requests */ +for (i =0 ; i <2 ; i++) { +if (must_wait[i]) { +continue ; + } +qemu_co_enter_next (&bs->throttled_reqs[i]); + } } -bool bdrv_io_limits_enabled (BlockDriverState *bs) +/* should be called before bdrv_set_io_limits if a limit is set */ +void bdrv_io_limits_enable (BlockDriverState *bs) { -BlockIOLimit *io_limits = &bs->io_limits; -return io_limits->bps[BLOCK_IO_LIMIT_READ ] - || io_limits->bps[BLOCK_IO_LIMIT_WRITE ] - || io_limits->bps[BLOCK_IO_LIMIT_TOTAL ] - || io_limits->iops[BLOCK_IO_LIMIT_READ ] - || io_limits->iops[BLOCK_IO_LIMIT_WRITE ] - || io_limits->iops[BLOCK_IO_LIMIT_TOTAL ]; +throttle_init (&bs->throttle_state, vm_clock, bdrv_throttle_timer_cb, bs); +qemu_co_queue_init (&bs->throttled_reqs[0 ]); +qemu_co_queue_init (&bs->throttled_reqs[1 ]); + bs->io_limits_enabled = true; }static void bdrv_io_limits_intercept (BlockDriverState *bs, -bool is_write,int nb_sectors) +bool is_write, +int nb_sectors) { -int64_t wait_time =-1 ; - -if (!qemu_co_queue_empty (&bs->throttled_reqs)) { -qemu_co_queue_wait (&bs->throttled_reqs); +if (!qemu_co_queue_empty (&bs->throttled_reqs[is_write])) { +qemu_co_queue_wait (&bs->throttled_reqs[is_write]); }/* In fact, we hope to keep each request's timing, in FIFO mode. The next
@@ -181,13 +187,11 @@ static void bdrv_io_limits_intercept(BlockDriverState *bs,* be still in throttled_reqs queue. */ -while (bdrv_exceed_io_limits (bs, nb_sectors, is_write, &wait_time)) { -qemu_mod_timer (bs->block_timer, - wait_time +qemu_get_clock_ns (vm_clock)); -qemu_co_queue_wait_insert_head (&bs->throttled_reqs); +while (throttle_start (&bs->throttle_state, is_write, nb_sectors)) { +qemu_co_queue_wait_insert_head (&bs->throttled_reqs[is_write]); } -qemu_co_queue_next (&bs->throttled_reqs); +qemu_co_queue_next (&bs->throttled_reqs[is_write]); }/* check if the path starts with "<protocol>:" */
@@ -1106,11 +1110,6 @@ int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,bdrv_dev_change_media_cb (bs, true); } -/* throttling disk I/O limits */ -if (bs->io_limits_enabled) { -bdrv_io_limits_enable (bs); - } -return 0 ; unlink_and_fail:
@@ -1446,16 +1445,15 @@ void bdrv_drain_all(void)* a busy wait. */ QTAILQ_FOREACH(bs, &bdrv_states, list) { -while (qemu_co_enter_next (&bs->throttled_reqs)) { - busy = true; - } + busy =bdrv_drain_throttled (bs); } }while (busy);/* If requests are still pending there is a bug somewhere */ QTAILQ_FOREACH(bs, &bdrv_states, list) {assert (QLIST_EMPTY (&bs->tracked_requests)); -assert (qemu_co_queue_empty (&bs->throttled_reqs)); +assert (qemu_co_queue_empty (&bs->throttled_reqs[0 ])); +assert (qemu_co_queue_empty (&bs->throttled_reqs[1 ])); } }
@@ -1491,13 +1489,12 @@ static void bdrv_move_feature_fields(BlockDriverState *bs_dest, bs_dest->enable_write_cache = bs_src->enable_write_cache; -/* i/o timing parameters */ - bs_dest->slice_start = bs_src->slice_start; - bs_dest->slice_end = bs_src->slice_end; - bs_dest->slice_submitted = bs_src->slice_submitted; - bs_dest->io_limits = bs_src->io_limits; - bs_dest->throttled_reqs = bs_src->throttled_reqs; - bs_dest->block_timer = bs_src->block_timer; +/* i/o throttled req */ +memcpy (&bs_dest->throttle_state, + &bs_src->throttle_state, +sizeof (ThrottleState)); + bs_dest->throttled_reqs[0 ] = bs_src->throttled_reqs[0 ]; + bs_dest->throttled_reqs[1 ] = bs_src->throttled_reqs[1 ]; bs_dest->io_limits_enabled = bs_src->io_limits_enabled;/* r/w error */
@@ -1544,7 +1541,7 @@ void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old)assert (bs_new->dev ==NULL );assert (bs_new->in_use ==0 );assert (bs_new->io_limits_enabled == false); -assert (bs_new->block_timer ==NULL ); +assert (!throttle_have_timer (&bs_new->throttle_state)); tmp = *bs_new; *bs_new = *bs_old;
@@ -1563,7 +1560,7 @@ void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old)assert (bs_new->job ==NULL );assert (bs_new->in_use ==0 );assert (bs_new->io_limits_enabled == false); -assert (bs_new->block_timer ==NULL ); +assert (!throttle_have_timer (&bs_new->throttle_state));bdrv_rebind (bs_new);bdrv_rebind (bs_old);
@@ -1854,8 +1851,15 @@ int bdrv_commit_all(void)* * This function should be called when a tracked request is completing. */ -static void tracked_request_end (BdrvTrackedRequest *req) +static void tracked_request_end (BlockDriverState *bs, +BdrvTrackedRequest *req, +bool is_write) { +/* throttling disk I/O */ +if (bs->io_limits_enabled) { +throttle_end (&bs->throttle_state, is_write); + } +QLIST_REMOVE (req, list);qemu_co_queue_restart_all (&req->wait_queue); }
@@ -1868,6 +1872,11 @@ static void tracked_request_begin(BdrvTrackedRequest *req,int64_t sector_num, int nb_sectors, bool is_write) { + /* throttling disk I/O */ + if (bs->io_limits_enabled) { +bdrv_io_limits_intercept (bs ,is_write ,nb_sectors ); + } + *req = (BdrvTrackedRequest){ .bs = bs, .sector_num = sector_num,
@@ -2506,11 +2515,6 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,return -EIO ; } -/* throttling disk read I/O */ -if (bs->io_limits_enabled) { -bdrv_io_limits_intercept (bs, false, nb_sectors); - } -if (bs->copy_on_read) { flags |=BDRV_REQ_COPY_ON_READ ; }
@@ -2541,7 +2545,7 @@ static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs, ret = drv->bdrv_co_readv (bs, sector_num, nb_sectors, qiov); out: -tracked_request_end (&req); +tracked_request_end (bs, &req, false);if (flags &BDRV_REQ_COPY_ON_READ ) { bs->copy_on_read_in_flight--;
@@ -2619,11 +2623,6 @@ static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,return -EIO ; } -/* throttling disk write I/O */ -if (bs->io_limits_enabled) { -bdrv_io_limits_intercept (bs, true, nb_sectors); - } -if (bs->copy_on_read_in_flight) {wait_for_overlapping_requests (bs, sector_num, nb_sectors); }
@@ -2652,7 +2651,7 @@ static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs, bs->wr_highest_sector = sector_num + nb_sectors -1 ; } -tracked_request_end (&req); +tracked_request_end (bs, &req, true);return ret; }
@@ -2745,14 +2744,6 @@ void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr) *nb_sectors_ptr = length; } -/* throttling disk io limits */ -void bdrv_set_io_limits (BlockDriverState *bs, -BlockIOLimit *io_limits) -{ - bs->io_limits = *io_limits; - bs->io_limits_enabled =bdrv_io_limits_enabled (bs); -} -void bdrv_set_on_error (BlockDriverState *bs,BlockdevOnError on_read_error,BlockdevOnError on_write_error) {
@@ -3562,169 +3553,6 @@ void bdrv_aio_cancel(BlockDriverAIOCB *acb) acb->aiocb_info->cancel (acb); } -/* block I/O throttling */ -static bool bdrv_exceed_bps_limits (BlockDriverState *bs,int nb_sectors, -bool is_write,double elapsed_time,uint64_t *wait) -{ -uint64_t bps_limit =0 ; -uint64_t extension; -double bytes_limit, bytes_base, bytes_res; -double slice_time, wait_time; - -if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL ]) { - bps_limit = bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL ]; - }else if (bs->io_limits.bps[is_write]) { - bps_limit = bs->io_limits.bps[is_write]; - }else { -if (wait) { - *wait =0 ; - } - -return false; - } - - slice_time = bs->slice_end - bs->slice_start; - slice_time /= (NANOSECONDS_PER_SECOND ); - bytes_limit = bps_limit * slice_time; - bytes_base = bs->slice_submitted.bytes[is_write]; -if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL ]) { - bytes_base += bs->slice_submitted.bytes[!is_write]; - } - -/* bytes_base: the bytes of data which have been read/written; and -* it is obtained from the history statistic info. -* bytes_res: the remaining bytes of data which need to be read/written. -* (bytes_base + bytes_res) / bps_limit: used to calcuate -* the total time for completing reading/writting all data. -*/ - bytes_res = (unsigned ) nb_sectors *BDRV_SECTOR_SIZE ; - -if (bytes_base + bytes_res <= bytes_limit) { -if (wait) { - *wait =0 ; - } - -return false; - } - -/* Calc approx time to dispatch */ - wait_time = (bytes_base + bytes_res) / bps_limit - elapsed_time; - -/* When the I/O rate at runtime exceeds the limits, -* bs->slice_end need to be extended in order that the current statistic -* info can be kept until the timer fire, so it is increased and tuned -* based on the result of experiment. -*/ - extension = wait_time *NANOSECONDS_PER_SECOND ; - extension =DIV_ROUND_UP (extension,BLOCK_IO_SLICE_TIME ) * -BLOCK_IO_SLICE_TIME ; - bs->slice_end += extension; -if (wait) { - *wait = wait_time *NANOSECONDS_PER_SECOND ; - } - -return true; -} - -static bool bdrv_exceed_iops_limits (BlockDriverState *bs,bool is_write, -double elapsed_time,uint64_t *wait) -{ -uint64_t iops_limit =0 ; -double ios_limit, ios_base; -double slice_time, wait_time; - -if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL ]) { - iops_limit = bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL ]; - }else if (bs->io_limits.iops[is_write]) { - iops_limit = bs->io_limits.iops[is_write]; - }else { -if (wait) { - *wait =0 ; - } - -return false; - } - - slice_time = bs->slice_end - bs->slice_start; - slice_time /= (NANOSECONDS_PER_SECOND ); - ios_limit = iops_limit * slice_time; - ios_base = bs->slice_submitted.ios[is_write]; -if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL ]) { - ios_base += bs->slice_submitted.ios[!is_write]; - } - -if (ios_base +1 <= ios_limit) { -if (wait) { - *wait =0 ; - } - -return false; - } - -/* Calc approx time to dispatch, in seconds */ - wait_time = (ios_base +1 ) / iops_limit; -if (wait_time > elapsed_time) { - wait_time = wait_time - elapsed_time; - }else { - wait_time =0 ; - } - -/* Exceeded current slice, extend it by another slice time */ - bs->slice_end +=BLOCK_IO_SLICE_TIME ; -if (wait) { - *wait = wait_time *NANOSECONDS_PER_SECOND ; - } - -return true; -} - -static bool bdrv_exceed_io_limits (BlockDriverState *bs,int nb_sectors, -bool is_write,int64_t *wait) -{ -int64_t now, max_wait; -uint64_t bps_wait =0 , iops_wait =0 ; -double elapsed_time; -int bps_ret, iops_ret; - - now =qemu_get_clock_ns (vm_clock); -if (now > bs->slice_end) { - bs->slice_start = now; - bs->slice_end = now +BLOCK_IO_SLICE_TIME ; -memset (&bs->slice_submitted,0 ,sizeof (bs->slice_submitted)); - } - - elapsed_time = now - bs->slice_start; - elapsed_time /= (NANOSECONDS_PER_SECOND ); - - bps_ret =bdrv_exceed_bps_limits (bs, nb_sectors, - is_write, elapsed_time, &bps_wait); - iops_ret =bdrv_exceed_iops_limits (bs, is_write, - elapsed_time, &iops_wait); -if (bps_ret || iops_ret) { - max_wait = bps_wait > iops_wait ? bps_wait : iops_wait; -if (wait) { - *wait = max_wait; - } - - now =qemu_get_clock_ns (vm_clock); -if (bs->slice_end < now + max_wait) { - bs->slice_end = now + max_wait; - } - -return true; - } - -if (wait) { - *wait =0 ; - } - - bs->slice_submitted.bytes[is_write] += (int64_t )nb_sectors * -BDRV_SECTOR_SIZE ; - bs->slice_submitted.ios[is_write]++; - -return false; -} -/**************************************************************/ /* async block device emulation */
block/qapi.c +9 -12
@@ -223,18 +223,15 @@ void bdrv_query_info(BlockDriverState *bs, info->inserted->backing_file_depth =bdrv_get_backing_file_depth (bs);if (bs->io_limits_enabled) { - info->inserted->bps = - bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL ]; - info->inserted->bps_rd = - bs->io_limits.bps[BLOCK_IO_LIMIT_READ ]; - info->inserted->bps_wr = - bs->io_limits.bps[BLOCK_IO_LIMIT_WRITE ]; - info->inserted->iops = - bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL ]; - info->inserted->iops_rd = - bs->io_limits.iops[BLOCK_IO_LIMIT_READ ]; - info->inserted->iops_wr = - bs->io_limits.iops[BLOCK_IO_LIMIT_WRITE ]; +ThrottleConfig cfg; +throttle_get_config (&bs->throttle_state, &cfg); + info->inserted->bps = cfg.buckets[THROTTLE_BPS_TOTAL ].ups; + info->inserted->bps_rd = cfg.buckets[THROTTLE_BPS_READ ].ups; + info->inserted->bps_wr = cfg.buckets[THROTTLE_BPS_WRITE ].ups; + + info->inserted->iops = cfg.buckets[THROTTLE_OPS_TOTAL ].ups; + info->inserted->iops_rd = cfg.buckets[THROTTLE_OPS_READ ].ups; + info->inserted->iops_wr = cfg.buckets[THROTTLE_OPS_WRITE ].ups; } bs0 = bs;
blockdev.c +62 -53
162 lines not on this page · raw message
include/block/block.h +0 -1
@@ -107,7 +107,6 @@ void bdrv_info_stats(Monitor *mon, QObject **ret_data);/* disk I/O throttling */ void bdrv_io_limits_enable (BlockDriverState *bs);void bdrv_io_limits_disable (BlockDriverState *bs); -bool bdrv_io_limits_enabled (BlockDriverState *bs);void bdrv_init (void );void bdrv_init_with_whitelist (void );
include/block/block_int.h +7 -26
@@ -34,18 +34,12 @@#include "monitor/monitor.h" #include "qemu/hbitmap.h" #include "block/snapshot.h" +#include "qemu/throttle.h" #define BLOCK_FLAG_ENCRYPT 1#define BLOCK_FLAG_COMPAT6 4#define BLOCK_FLAG_LAZY_REFCOUNTS 8 -#define BLOCK_IO_LIMIT_READ 0 -#define BLOCK_IO_LIMIT_WRITE 1 -#define BLOCK_IO_LIMIT_TOTAL 2 - -#define BLOCK_IO_SLICE_TIME 100000000 -#define NANOSECONDS_PER_SECOND 1000000000.0 -#define BLOCK_OPT_SIZE "size"#define BLOCK_OPT_ENCRYPT "encryption"#define BLOCK_OPT_COMPAT6 "compat6"
@@ -69,17 +63,6 @@ typedef struct BdrvTrackedRequest {CoQueue wait_queue;/* coroutines blocked on this request */ } BdrvTrackedRequest; - -typedef struct BlockIOLimit { -int64_t bps[3 ]; -int64_t iops[3 ]; -}BlockIOLimit ; - -typedef struct BlockIOBaseValue { -uint64_t bytes[2 ]; -uint64_t ios[2 ]; -}BlockIOBaseValue ; -struct BlockDriver {const char *format_name;int instance_size;
@@ -263,13 +246,10 @@ struct BlockDriverState {/* number of in-flight copy-on-read requests */ unsigned int copy_on_read_in_flight; -/* the time for latest disk I/O */ -int64_t slice_start; -int64_t slice_end; -BlockIOLimit io_limits; -BlockIOBaseValue slice_submitted; -CoQueue throttled_reqs; -QEMUTimer *block_timer; +/* I/O throttling */ +ThrottleState throttle_state; +/* two throttled request queue so read don't starve write nor the reverse */ +CoQueue throttled_reqs[2 ];bool io_limits_enabled;/* I/O stats (display with "info blockstats"). */
@@ -308,7 +288,8 @@ struct BlockDriverState {int get_tmp_filename (char *filename,int size);void bdrv_set_io_limits (BlockDriverState *bs, -BlockIOLimit *io_limits); +ThrottleConfig *cfg); +/** * bdrv_add_before_write_notifier:
Signature · 1 line
1.7.10.4
Quoted 11 lines · Benoît Canet
This patchset implement continous leaky bucket throttling.
It works mostly on the general case.
The exception is where the load is composed of both reads and writes and two limitsiops_rdandiops_wrare set.
The resulting iops are a little above half of the given limits. I tried various strategies to avoid this: two timer, two throttled request queues or even a different algorithm using a priority queue.
The problem is still the same in every version of the code: reads and writes operation seems entangled.
Quoted 2 lines
Benoît Canet (2):
throttle: Add a new throttling API implementing continuus leaky
s/continuus/continuous/
Quoted 19 lines
bucket.
block: Enable the new throttling code in the block layer.block.c | 316 ++++++++------------------------ block/qapi.c | 21 +-- blockdev.c | 115 ++++++------ include/block/block.h | 1 - include/block/block_int.h | 33 +--- include/qemu/throttle.h | 111 ++++++++++++ util/Makefile.objs | 1 + util/throttle.c | 436 +++++++++++++++++++++++++++++++++++++++++++++ 8 files changed, 698 insertions(+), 336 deletions(-) create mode 100644 include/qemu/throttle.h create mode 100644 util/throttle.cSignature · 1 line
1.7.10.4
Signature · 1 line
Fam
Quoted 16 lines · Benoît Canet
This patchset implement continous leaky bucket throttling.
It works mostly on the general case.
The exception is where the load is composed of both reads and writes and two limitsiops_rdandiops_wrare set.
The resulting iops are a little above half of the given limits. I tried various strategies to avoid this: two timer, two throttled request queues or even a different algorithm using a priority queue.
The problem is still the same in every version of the code: reads and writes operation seems entangled.Benoît Canet (2):
throttle: Add a new throttling API implementing continuus leaky
bucket.
block: Enable the new throttling code in the block layer.
Quoted 11 lines
block.c | 316 ++++++++------------------------ block/qapi.c | 21 +-- blockdev.c | 115 ++++++------ include/block/block.h | 1 - include/block/block_int.h | 33 +--- include/qemu/throttle.h | 111 ++++++++++++ util/Makefile.objs | 1 + util/throttle.c | 436 +++++++++++++++++++++++++++++++++++++++++++++ 8 files changed, 698 insertions(+), 336 deletions(-) create mode 100644 include/qemu/throttle.h create mode 100644 util/throttle.c
I saw more discussion on IRC. Does this mean you will send another revision to address outstanding issues?
Just wanted to check if you are waiting for code review or if you are still developing the next patch revision.
Stefan
Quoted 5 lines
I saw more discussion on IRC. Does this mean you will send another revision to address outstanding issues?
Just wanted to check if you are waiting for code review or if you are still developing the next patch revision.
I am currently finishing to write unit tests for the next patch revision. Once unit tests are done I will rebase the previous QMP throttling patches on top of the new code then post the new series.
I think that the result will be a net improvement.
Thanks for asking
Best regards
Benoît