mm/damon: auto-tune aggregation interval
Subsystems (4)
SeongJae Park
maintainerv2 cover mm/damon: auto-tune aggregation interval
DAMON requires time-consuming and repetitive aggregation interval tuning. Introduce a feature for automating it using a feedback loop that aims an amount of observed access events, like auto-exposing cameras.
Background: Access Frequency Monitoring and Aggregation Interval
================================================================
DAMON checks if each memory element (damon_region) is accessed or not for every user-specified time interval called 'sampling interval'. It aggregates the check intervals on per-element counter called 'nr_accesses'. DAMON users can read the counters to get the access temperature of a given element. The counters are reset for every another user-specified time interval called 'aggregation interval'.
This can be illustrated as DAMON continuously capturing a snapshot of access events that happen and captured within the last aggregation interval. This implies the aggregation interval plays a key role for the quality of the snapshots, like the camera exposure time. If it is too short, the amount of access events that happened and captured for each snapshot is small, so each snapshot will show no many interesting things but just a cold and dark world with hopefuly one pale blue dot or two. If it is too long, too many events are aggregated in a single shot, so each snapshot will look like world of flames, or Muspellheim. It will be difficult to find practical insights in both cases.
Problem: Time Consuming and Repetitive Tuning
=============================================
The appropriate length of the aggregation interval depends on how frequently the system and workloads are making access events that DAMON can observe. Hence, users have to tune the interval with excessive amount of tests with the target system and workloads. If the system and workloads are changed, the tuning should be done again. If the characteristic of the workloads is dynamic, it becomes more challenging. It is therefore time-consuming and repetitive.
The tuning challenge mainly stems from the wrong question. It is not asking users what quality of monitoring results they want, but how DAMON should operate for their hidden goal. To make the right answer, users need to fully understand DAMON's mechanisms and the characteristics of their workloads. Users shouldn't be asked to understand the underlying mechanism. Understanding the characteristics of the workloads shouldn't be the role of users but DAMON.
Aim-oriented Feedback-driven Auto-Tuning
=========================================
Fortunately, the appropriate length of the aggregation interval can be inferred using a feedback loop. If the current snapshots are showing no much intresting information, in other words, if it shows only rare access events, increasing the aggregation interval helps, and vice versa. We tested this theory on a few real-world workloads, and documented one of the experience with an official DAMON monitoring intervals tuning guideline. Since it is a simple theory that requires repeatable tries, it can be a good job for machines.
Based on the guideline's theory, we design an automation of aggregation interval tuning, in a way similar to that of camera auto-exposure feature. It defines the amount of interesting information as the ratio of DAMON-observed access events that DAMON actually observed to theoretical maximum amount of it within each snapshot. Events are accounted in byte and sampling attempts granularity. For example, let's say there is a region of 'X' bytes size. DAMON tried access check smapling for the region 'Y' times in total for a given aggregation. Among the 'Y' attempts, 'Z' times it shown positive results. Then, the theoritical maximum number of access events for the region is 'X * Y'. And the number of access events that DAMON has observed for the region is 'X * Z'. The abount of the interesting information is '(X * Z / X * Y)'. Note that each snapshot would have multiple regions.
Users can set an arbitrary value of the ratio as their target. Once the target is set, the automation periodically measures the current value of the ratio and increase or decrease the aggregation interval if the ratio value is lower or higher than the target. The amount of the change is proportion to the distance between the current adn the target values.
To avoid auto-tuning goes too long way, let users set the minimum and the maximum aggregation interval times. Changing only aggregation interval while sampling interval is kept makes the maximum level of access frequency in each snapshot, or discernment of regions inconsistent. Also, unnecessarily short sampling interval causes meaningless monitoring overhed. The automation therefore adjusts the sampling interval together with aggregation interval, while keeping the ratio between the two intervals. Users can set the ratio, or the discernment.
Discussion
==========
The modified question (aimed amount of access events, or lights, in each snapshot) is easy to answer by both the users and the kernel. If users are interested in finding more cold regions, the value should be lower, and vice versa. If users have no idea, kernel can suggest a fair default value based on some theories and experiments. For example, based on the Pareto principle (80/20 rule), we could expect 20% target ratio will capture 80% of real access events. Since 80% might be too high, applying the rule once again, 4% (20% * 20%) may capture about 56% (80% * 80%) of real access events.
Sampling to aggregation intervals ratio and min/max aggregation intervals are also arguably easy to answer. What users want is discernment of regions for efficient system operation, for examples, X amount of colder regions or Y amount of warmer regions, not exactly how many times each cache line is accessed in nanoseconds degree. The appropriate min/max aggregation interval can relatively naively set, and may better to set for aimed monitoring overhead. Since sampling interval is directly deciding the overhead, setting it based on the sampling interval can be easy. With my experiences, I'd argue the intervals ratio 0.05, and 5 milliseconds to 20 seconds sampling interval range (100 milliseconds to 400 seconds aggregation interval) can be a good default suggestion.
Evaluation
==========
We confirmed the tuning works as expected with a few simple workloads including kernel builds and an in-memory caching representative benchmark[1]. We will conduct more evaluations with more workloads and share the results with more details by the time that we drop the RFC tag.
Changelog
=========
Changes from RFC v1
(https://lore.kernel.org/[email protected])
- Replace the target metric from positive samples ratio to DAMON-observed access samples ratio
- Fix wrong max events accounting bug
- Fix double-increase of next_aggregation_sis
SeongJae Park (8):
mm/damon: add data structure for monitoring intervals auto-tuning mm/damon/core: implement intervals auto-tuning mm/damon/sysfs: implement intervals tuning goal directory mm/damon/sysfs: commit intervals tuning goal mm/damon/sysfs: implement a command to update auto-tuned monitoring
intervals
Docs/mm/damon/design: document for intervals auto-tuning Docs/ABI/damon: document intervals auto-tuning ABI Docs/admin-guide/mm/damon/usage: add intervals_goal directory on the
hierarchy
.../ABI/testing/sysfs-kernel-mm-damon | 30 +++ Documentation/admin-guide/mm/damon/usage.rst | 25 ++ Documentation/mm/damon/design.rst | 50 ++++ include/linux/damon.h | 43 ++++ mm/damon/core.c | 98 ++++++++ mm/damon/sysfs.c | 216 ++++++++++++++++++ 6 files changed, 462 insertions(+)
base-commit: 9e7d9145ab8ce407acc540fc29133c471bc29046
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 1/8 mm/damon: add data structure for monitoring intervals auto-tuning
Add data structures for DAMON sampling and aggregation intervals automatic tuning that aims specific amount of DAMON-observed access events per snapshot. In more detail, define the data structure for the tuning goal, link it to the monitoring attributes data structure so that DAMON kernel API callers can make the request, and update parameters setup DAMON function to respect the new parameter.
--- include/linux/damon.h | 27 +++++++++++++++++++++++++++ mm/damon/core.c | 22 ++++++++++++++++++++++ 2 files changed, 49 insertions(+)
include/linux/damon.h +27 -0
@@ -664,12 +664,38 @@ struct damon_call_control {bool canceled; }; +/** +* struct damon_intervals_goal - Monitoring intervals auto-tuning goal. +* +* @access_bp: Access events observation ratio to achieve in bp. +* @aggrs: Number of aggregations to acheive @access_bp within. +* @min_sample_us: Minimum resulting sampling interval in microseconds. +* @max_sample_us: Maximum resulting sampling interval in microseconds. +* +* DAMON automatically tunes &damon_attrs->sample_interval and +* &damon_attrs->aggr_interval aiming the ratio in bp (1/10,000) of +* DAMON-observed access events to theoretical maximum amount within @aggrs +* aggregations be same to @access_bp. The logic increases +* &damon_attrs->aggr_interval and &damon_attrs->sampling_interval in same +* ratio if the current access events observation ratio is lower than the +* target for each @aggrs aggregations, and vice versa. +* +* If @aggrs is zero, the tuning is disabled and hence this struct is ignored. +*/ +struct damon_intervals_goal { +unsigned long access_bp; +unsigned long aggrs; +unsigned long min_sample_us; +unsigned long max_sample_us; +}; +/** * struct damon_attrs - Monitoring attributes for accuracy/overhead control. * * @sample_interval: The time between access samplings. * @aggr_interval: The time between monitor results aggregations. * @ops_update_interval: The time between monitoring operations updates. +* @intervals_goal: Intervals auto-tuning goal. * @min_nr_regions: The minimum number of adaptive monitoring * regions. * @max_nr_regions: The maximum number of adaptive monitoring
@@ -689,6 +715,7 @@ struct damon_attrs {unsigned long sample_interval;unsigned long aggr_interval;unsigned long ops_update_interval; +struct damon_intervals_goal intervals_goal;unsigned long min_nr_regions;unsigned long max_nr_regions; };
mm/damon/core.c +22 -0
@@ -638,6 +638,25 @@ static void damon_update_monitoring_results(struct damon_ctx *ctx, r, old_attrs, new_attrs); } +/* +* damon_valid_intervals_goal() - return if the intervals goal of @attrs is +* valid. +*/ +static bool damon_valid_intervals_goal (struct damon_attrs *attrs) +{ +struct damon_intervals_goal *goal = &attrs->intervals_goal; + +/* tuning is disabled */ +if (!goal->aggrs) +return true; +if (goal->min_sample_us > goal->max_sample_us) +return false; +if (attrs->sample_interval < goal->min_sample_us || + goal->max_sample_us < attrs->sample_interval) +return false; +return true; +} +/** * damon_set_attrs() - Set attributes for the monitoring. * @ctx: monitoring context
@@ -658,6 +677,9 @@ int damon_set_attrs(struct damon_ctx *ctx, struct damon_attrs *attrs) attrs->sample_interval : 1;struct damos *s; +if (!damon_valid_intervals_goal (attrs)) +return -EINVAL ; +if (attrs->min_nr_regions <3 )return -EINVAL ;if (attrs->min_nr_regions > attrs->max_nr_regions)
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 2/8 mm/damon/core: implement intervals auto-tuning
In reply to SeongJae ParkImplement the DAMON sampling and aggregation intervals auto-tuning mechanism as briefly described on 'struct damon_intervals_goal'. The core part for deciding the direction and amount of the changes is implemented reusing the feedback loop function which is being used for DAMOS quotas auto-tuning. Unlike the DAMOS quotas auto-tuning use case, limit the maximum decreasing amount after the adjustment to 50% of the current value, though. This is because the intervals have no good merits at rapid reductions since it could unnecessarily increase the monitoring overhead.
--- include/linux/damon.h | 16 +++++++++ mm/damon/core.c | 76 +++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 92 insertions(+)
include/linux/damon.h +16 -0
@@ -718,6 +718,17 @@ struct damon_attrs {struct damon_intervals_goal intervals_goal;unsigned long min_nr_regions;unsigned long max_nr_regions; +/* private: internal use only */ +/* +* @aggr_interval to @sample_interval ratio. +* Core-external components call damon_set_attrs() with &damon_attrs +* that this field is unset. In the case, damon_set_attrs() sets this +* field of resulting &damon_attrs. Core-internal components such as +* kdamond_tune_intervals() calls damon_set_attrs() with &damon_attrs +* that this field is set. In the case, damon_set_attrs() just keep +* it. +*/ +unsigned long aggr_samples; };/**
@@ -766,6 +777,11 @@ struct damon_ctx {* update */ unsigned long next_ops_update_sis; +/* +* number of sample intervals that should be passed before next +* intervals tuning +*/ +unsigned long next_intervals_tune_sis;/* for waiting until the execution of the kdamond_fn is started */ struct completion kdamond_started;/* for scheme quotas prioritization */
mm/damon/core.c +76 -0
100 lines not on this page · raw message
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 3/8 mm/damon/sysfs: implement intervals tuning goal directory
In reply to SeongJae ParkImplement DAMON sysfs interface directory and its files for setting DAMON sampling and aggregation intervals auto-tuning goal.
--- mm/damon/sysfs.c | 189 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 189 insertions(+)
mm/damon/sysfs.c +189 -0
@@ -408,6 +408,164 @@ static const struct kobj_type damon_sysfs_targets_ktype = { .default_groups = damon_sysfs_targets_groups, }; +/* +* intervals goal directory +*/ + +struct damon_sysfs_intervals_goal { +struct kobject kobj; +unsigned long access_bp; +unsigned long aggrs; +unsigned long min_sample_us; +unsigned long max_sample_us; +}; + +static struct damon_sysfs_intervals_goal *damon_sysfs_intervals_goal_alloc ( +unsigned long access_bp,unsigned long aggrs, +unsigned long min_sample_us,unsigned long max_sample_us) +{ +struct damon_sysfs_intervals_goal *goal =kmalloc (sizeof (*goal), +GFP_KERNEL ); + +if (!goal) +return NULL ; + + goal->kobj = (struct kobject ){}; + goal->access_bp = access_bp; + goal->aggrs = aggrs; + goal->min_sample_us = min_sample_us; + goal->max_sample_us = max_sample_us; +return goal; +} + +static ssize_t access_bp_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); + +return sysfs_emit (buf,"%lu\n" , goal->access_bp); +} + +static ssize_t access_bp_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); +unsigned long nr; +int err =kstrtoul (buf,0 , &nr); + +if (err) +return err; + + goal->access_bp = nr; +return count; +} + +static ssize_t aggrs_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); + +return sysfs_emit (buf,"%lu\n" , goal->aggrs); +} + +static ssize_t aggrs_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); +unsigned long nr; +int err =kstrtoul (buf,0 , &nr); + +if (err) +return err; + + goal->aggrs = nr; +return count; +} + +static ssize_t min_sample_us_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); + +return sysfs_emit (buf,"%lu\n" , goal->min_sample_us); +} + +static ssize_t min_sample_us_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); +unsigned long nr; +int err =kstrtoul (buf,0 , &nr); + +if (err) +return err; + + goal->min_sample_us = nr; +return count; +} + +static ssize_t max_sample_us_show (struct kobject *kobj, +struct kobj_attribute *attr,char *buf) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); + +return sysfs_emit (buf,"%lu\n" , goal->max_sample_us); +} + +static ssize_t max_sample_us_store (struct kobject *kobj, +struct kobj_attribute *attr,const char *buf,size_t count) +{ +struct damon_sysfs_intervals_goal *goal =container_of (kobj, + struct damon_sysfs_intervals_goal, kobj); +unsigned long nr; +int err =kstrtoul (buf,0 , &nr); + +if (err) +return err; + + goal->max_sample_us = nr; +return count; +} + +static void damon_sysfs_intervals_goal_release (struct kobject *kobj) +{ +kfree (container_of(kobj ,struct damon_sysfs_intervals_goal ,kobj )); +} + +static struct kobj_attribute damon_sysfs_intervals_goal_access_bp_attr = +__ATTR_RW_MODE (access_bp,0600 ); + +static struct kobj_attribute damon_sysfs_intervals_goal_aggrs_attr = +__ATTR_RW_MODE (aggrs,0600 ); + +static struct kobj_attribute damon_sysfs_intervals_goal_min_sample_us_attr = +__ATTR_RW_MODE (min_sample_us,0600 ); + +static struct kobj_attribute damon_sysfs_intervals_goal_max_sample_us_attr = +__ATTR_RW_MODE (max_sample_us,0600 ); + +static struct attribute *damon_sysfs_intervals_goal_attrs[] = { + &damon_sysfs_intervals_goal_access_bp_attr.attr, + &damon_sysfs_intervals_goal_aggrs_attr.attr, + &damon_sysfs_intervals_goal_min_sample_us_attr.attr, + &damon_sysfs_intervals_goal_max_sample_us_attr.attr, +NULL , +}; +ATTRIBUTE_GROUPS (damon_sysfs_intervals_goal); + +static const struct kobj_type damon_sysfs_intervals_goal_ktype = { + .release = damon_sysfs_intervals_goal_release, + .sysfs_ops = &kobj_sysfs_ops, + .default_groups = damon_sysfs_intervals_goal_groups, +}; +/* * intervals directory */
@@ -417,6 +575,7 @@ struct damon_sysfs_intervals {unsigned long sample_us;unsigned long aggr_us;unsigned long update_us; +struct damon_sysfs_intervals_goal *intervals_goal; }; static struct damon_sysfs_intervals *damon_sysfs_intervals_alloc(
@@ -436,6 +595,32 @@ static struct damon_sysfs_intervals *damon_sysfs_intervals_alloc(return intervals; } +static int damon_sysfs_intervals_add_dirs (struct damon_sysfs_intervals *intervals) +{ +struct damon_sysfs_intervals_goal *goal; +int err; + + goal =damon_sysfs_intervals_goal_alloc (0 ,0 ,0 ,0 ); +if (!goal) +return -ENOMEM ; + + err =kobject_init_and_add (&goal->kobj, + &damon_sysfs_intervals_goal_ktype, &intervals->kobj, +"intervals_goal" ); +if (err) { +kobject_put (&goal->kobj); + intervals->intervals_goal =NULL ; +return err; + } + intervals->intervals_goal = goal; +return 0 ; +} + +static void damon_sysfs_intervals_rm_dirs (struct damon_sysfs_intervals *intervals) +{ +kobject_put (&intervals->intervals_goal->kobj); +} +static ssize_t sample_us_show (struct kobject *kobj,struct kobj_attribute *attr,char *buf) {
@@ -569,6 +754,9 @@ static int damon_sysfs_attrs_add_dirs(struct damon_sysfs_attrs *attrs) err =kobject_init_and_add (&intervals->kobj, &damon_sysfs_intervals_ktype, &attrs->kobj,"intervals" ); +if (err) + goto put_intervals_out; + err =damon_sysfs_intervals_add_dirs (intervals);if (err) goto put_intervals_out; attrs->intervals = intervals;
@@ -599,6 +787,7 @@ static int damon_sysfs_attrs_add_dirs(struct damon_sysfs_attrs *attrs)static void damon_sysfs_attrs_rm_dirs (struct damon_sysfs_attrs *attrs) {kobject_put (&attrs->nr_regions_range->kobj); +damon_sysfs_intervals_rm_dirs (attrs->intervals);kobject_put (&attrs->intervals->kobj); }
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 4/8 mm/damon/sysfs: commit intervals tuning goal
In reply to SeongJae ParkConnect DAMON sysfs interface for sampling and aggregation intervals auto-tuning with DAMON core API, so that users can really use the feature using the sysfs files.
--- mm/damon/sysfs.c | 7 +++++++ 1 file changed, 7 insertions(+)
mm/damon/sysfs.c +7 -0
@@ -1273,11 +1273,18 @@ static int damon_sysfs_set_attrs(struct damon_ctx *ctx, struct damon_sysfs_attrs *sys_attrs) {struct damon_sysfs_intervals *sys_intervals = sys_attrs->intervals; +struct damon_sysfs_intervals_goal *sys_goal = + sys_intervals->intervals_goal;struct damon_sysfs_ul_range *sys_nr_regions = sys_attrs->nr_regions_range;struct damon_attrs attrs = { .sample_interval = sys_intervals->sample_us, .aggr_interval = sys_intervals->aggr_us, + .intervals_goal = { + .access_bp = sys_goal->access_bp, + .aggrs = sys_goal->aggrs, + .min_sample_us = sys_goal->min_sample_us, + .max_sample_us = sys_goal->max_sample_us}, .ops_update_interval = sys_intervals->update_us, .min_nr_regions = sys_nr_regions->min, .max_nr_regions = sys_nr_regions->max,
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 5/8 mm/damon/sysfs: implement a command to update auto-tuned monitoring intervals
In reply to SeongJae ParkDAMON kernel API callers can show auto-tuned sampling and aggregation intervals from the monmitoring attributes data structure. That can be useful for debugging or tuning of the feature. DAMON user-space ABI users has no way to see that, though. Implement a new DAMON sysfs interface command, namely 'update_tuned_intervals', for the purpose. If the command is written to the kdamond state file, the tuned sampling and aggregation intervals will be updated to the corresponding sysfs interface files.
--- mm/damon/sysfs.c | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+)
mm/damon/sysfs.c +20 -0
@@ -1213,6 +1213,11 @@ enum damon_sysfs_cmd {* effective size quota of the scheme in bytes. */ DAMON_SYSFS_CMD_UPDATE_SCHEMES_EFFECTIVE_QUOTAS , +/* +* @DAMON_SYSFS_CMD_UPDATE_TUNED_INTERVALS: Update the tuned monitoring +* intevals. +*/ + DAMON_SYSFS_CMD_UPDATE_TUNED_INTERVALS,/* * @NR_DAMON_SYSFS_CMDS: Total number of DAMON sysfs commands. */
@@ -1230,6 +1235,7 @@ static const char * const damon_sysfs_cmd_strs[] = {"update_schemes_tried_regions" ,"clear_schemes_tried_regions" ,"update_schemes_effective_quotas" , + "update_tuned_intervals", };/*
@@ -1502,6 +1508,17 @@ static int damon_sysfs_upd_schemes_effective_quotas(void *data)return 0 ; } +static int damon_sysfs_upd_tuned_intervals (void *data) +{ +struct damon_sysfs_kdamond *kdamond = data; +struct damon_ctx *ctx = kdamond->damon_ctx; + + kdamond->contexts->contexts_arr[0 ]->attrs->intervals->sample_us = + ctx->attrs.sample_interval; + kdamond->contexts->contexts_arr[0 ]->attrs->intervals->aggr_us = + ctx->attrs.aggr_interval; +return 0 ; +}/* * damon_sysfs_cmd_request_callback() - DAMON callback for handling requests.
@@ -1723,6 +1740,9 @@ static int damon_sysfs_handle_cmd(enum damon_sysfs_cmd cmd,return damon_sysfs_damon_call ( damon_sysfs_upd_schemes_effective_quotas, kdamond); +case DAMON_SYSFS_CMD_UPDATE_TUNED_INTERVALS : +return damon_sysfs_damon_call ( + damon_sysfs_upd_tuned_intervals, kdamond);default :break ; }
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 6/8 Docs/mm/damon/design: document for intervals auto-tuning
In reply to SeongJae ParkDocument the design of DAMON sampling and aggregation intervals auto-tuning.
--- Documentation/mm/damon/design.rst | 46 +++++++++++++++++++++++++++++++ 1 file changed, 46 insertions(+)
Documentation/mm/damon/design.rst +46 -0
@@ -313,6 +313,10 @@ sufficient for the given purpose, it shouldn't be unnecessarily further lowered. It is recommended to be set proportional to ``aggregation interval``. By default, the ratio is set as ``1/20``, and it is still recommended. +Based on the manual tuning guide, DAMON provides more intuitive knob-based +intervals auto tuning mechanism. Please refer to :ref:`the design document of +the feature <damon_design_monitoring_intervals_autotuning>` for detail. + Refer to below documents for an example tuning based on the above guide. .. toctree::
@@ -321,6 +325,48 @@ Refer to below documents for an example tuning based on the above guide. monitoring_intervals_tuning_example +.. _damon_design_monitoring_intervals_autotuning: + +Monitoring Intervals Auto-tuning +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + +DAMON provides automatic tuning of the ``sampling interval`` and ``aggregation +interval`` based on the :ref:`the tuning guide idea +<damon_design_monitoring_params_tuning_guide>`. The tuning mechanism allows +users to set the aimed amount of access events to observe via DAMON within +given time interval. The target can be specified by the user as a ratio of +DAMON-observed access events to the theoretical maximum amount of the events +(``access_bp``) that measured within a given number of aggregations +(``aggrs``). + +The DAMON-observed access events are calculated in byte granularity based on +DAMON :ref:`region assumption <damon_design_region_based_sample>`. For +example, if a region of size ``X`` bytes of ``Y`` ``nr_accesses`` is found, it +means ``X * Y`` access events are observed by DAMON. Theoretical maximum +access events for the region is calculated in same way, but replacing ``Y`` +with theoretical maximum ``nr_accesses``, which can be calculated as +``aggregation interval / sampling interval``. + +The mechanism calculates the ratio of access events for ``aggrs`` aggregations, +and increases or decrease the ``sampleing interval`` and ``aggregation +interval`` in same ratio, if the observed access ratio is lower or higher than +the target, respectively. The ratio of the intervals change is decided in +proportion to the distance between current samples ratio and the target ratio. + +The user can further set the minimum and maximum ``sampling interval`` that can +be set by the tuning mechanism using two parameters (``min_sample_us`` and +``max_sample_us``). Because the tuning mechanism changes ``sampling interval`` +and ``aggregation interval`` in same ratio always, the minimum and maximum +``aggregation interval`` after each of the tuning changes can automatically set +together. + +The tuning is turned off by default, and need to be set explicitly by the user. +As a rule of thumbs and the Parreto principle, 4% access samples ratio target +is recommended. Note that Parreto principle (80/20 rule) has applied twice. +That is, assumes 4% (20% of 20%) DAMON-observed access events ratio (source) +to capture 64% (80% multipled by 80%) real access events (outcomes). + + .. _damon_design_damos: Operation Schemes
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 7/8 Docs/ABI/damon: document intervals auto-tuning ABI
In reply to SeongJae ParkDocument the DAMON user-space ABI for DAMON sampling and aggregation intervals auto-tuning.
--- .../ABI/testing/sysfs-kernel-mm-damon | 30 +++++++++++++++++++ 1 file changed, 30 insertions(+)
Documentation/ABI/testing/sysfs-kernel-mm-damon +30 -0
@@ -91,6 +91,36 @@ Description: Writing a value to this file sets the update interval of the DAMON context in microseconds as the value. Reading this file returns the value. +What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/intervals/intrvals_goal/access_bp +Date: Feb 2025 +Contact: SeongJae Park <[email protected]> +Description: Writing a value to this file sets the monitoring intervals + auto-tuning target DAMON-observed access events ratio within + the given time interval (aggrs in same directory), in bp + (1/10,000). Reading this file returns the value. + +What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/intervals/intrvals_goal/aggrs +Date: Feb 2025 +Contact: SeongJae Park <[email protected]> +Description: Writing a value to this file sets the time interval to achieve + the monitoring intervals auto-tuning target DAMON-observed + access events ratio (access_bp in same directory) within. + Reading this file returns the value. + +What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/intervals/intrvals_goal/min_sample_us +Date: Feb 2025 +Contact: SeongJae Park <[email protected]> +Description: Writing a value to this file sets the minimum value of + auto-tuned sampling interval in microseconds. Reading this + file returns the value. + +What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/intervals/intrvals_goal/max_sample_us +Date: Feb 2025 +Contact: SeongJae Park <[email protected]> +Description: Writing a value to this file sets the maximum value of + auto-tuned sampling interval in microseconds. Reading this + file returns the value. + What: /sys/kernel/mm/damon/admin/kdamonds/<K>/contexts/<C>/monitoring_attrs/nr_regions/min WDate: Mar 2022
Signature · 1 line
2.39.5
SeongJae Park
maintainerRFC PATCH v2 8/8 Docs/admin-guide/mm/damon/usage: add intervals_goal directory on the hierarchy
In reply to SeongJae ParkDocument DAMON sysfs interface usage for DAMON sampling and aggregation intervals auto-tuning.
--- Documentation/admin-guide/mm/damon/usage.rst | 25 ++++++++++++++++++++ Documentation/mm/damon/design.rst | 4 ++++ 2 files changed, 29 insertions(+)
Documentation/admin-guide/mm/damon/usage.rst +25 -0
@@ -64,6 +64,7 @@ comma (","). │ │ │ │ :ref:`0 <sysfs_context>`/avail_operations,operations │ │ │ │ │ :ref:`monitoring_attrs <sysfs_monitoring_attrs>`/ │ │ │ │ │ │ intervals/sample_us,aggr_us,update_us + │ │ │ │ │ │ │ intervals_goal/access_bp,aggrs,min_sample_us,max_sample_us │ │ │ │ │ │ nr_regions/min,max │ │ │ │ │ :ref:`targets <sysfs_targets>`/nr_targets │ │ │ │ │ │ :ref:`0 <sysfs_target>`/pid_target
@@ -132,6 +133,11 @@ Users can write below commands for the kdamond to the ``state`` file. - ``off``: Stop running. - ``commit``: Read the user inputs in the sysfs files except ``state`` file again. +- ``update_tuned_intervals``: Update the contents of ``sample_us`` and + ``aggr_us`` files of the kdamond with the auto-tuning applied ``sampling + interval`` and ``aggregation interval`` for the files. Please refer to + :ref:`intervals_goal section <damon_usage_sysfs_monitoring_intervals_goal>` + for more details. - ``commit_schemes_quota_goals``: Read the DAMON-based operation schemes' :ref:`quota goals <sysfs_schemes_quota_goals>`. - ``update_schemes_stats``: Update the contents of stats files for each
@@ -213,6 +219,25 @@ writing to and rading from the files. For more details about the intervals and monitoring regions range, please refer to the Design document (:doc:`/mm/damon/design`). +.. _damon_usage_sysfs_monitoring_intervals_goal: + +contexts/<N>/monitoring_attrs/intervals/intervals_goal/ +------------------------------------------------------- + +Under the ``intervals`` directory, one directory for automated tuning of +``sample_us`` and ``aggr_us``, namely ``intervals_goal`` directory also exists. +Under the directory, four files for the auto-tuning control, namely +``access_bp``, ``aggrs``, ``min_sample_us`` and ``max_sample_us`` exist. +Please refer to the :ref:`design document of the feature +<damon_design_monitoring_intervals_autotuning>` for the internal of the tuning +mechanism. Reading and writing the four files under ``intervals_goal`` +directory shows and updates the tuning parameters that described in the +:ref:design doc <damon_design_monitoring_intervals_autotuning>` with the same +names. The tuning starts with the user-set ``sample_us`` and ``aggr_us``. The +tuning-applied current values of the two intervals can be read from the +``sample_us`` and ``aggr_us`` files after writing ``update_tuned_intervals`` to +the ``state`` file. + .. _sysfs_targets: contexts/<N>/targets/
Documentation/mm/damon/design.rst +4 -0
@@ -366,6 +366,10 @@ is recommended. Note that Parreto principle (80/20 rule) has applied twice. That is, assumes 4% (20% of 20%) DAMON-observed access events ratio (source) to capture 64% (80% multipled by 80%) real access events (outcomes). +To know how user-space can use this feature via :ref:`DAMON sysfs interface +<sysfs_interface>`, refer to :ref:`intervals_goal <sysfs_scheme>` part of +the documentation. + .. _damon_design_damos:
Signature · 1 line
2.39.5
SeongJae Park
maintainerQuoted 4 lines · On Fri, 28 Feb 2025 14:03:20 -0800 SeongJae Park <[email protected]> wrote:
DAMON requires time-consuming and repetitive aggregation interval tuning. Introduce a feature for automating it using a feedback loop that aims an amount of observed access events, like auto-exposing cameras.
[...]
Quoted 6 lines
Evaluation
==========We confirmed the tuning works as expected with a few simple workloads including kernel builds and an in-memory caching representative benchmark[1].
Forgot adding the link to the benchmark, sorry. It is
https://github.com/facebookresearch/DCPerf/blob/main/packages/tao_bench/README.md
Thanks,
SJ
[...]