CN108763000A - A kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform - Google Patents
A kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform Download PDFInfo
- Publication number
- CN108763000A CN108763000A CN201810503061.8A CN201810503061A CN108763000A CN 108763000 A CN108763000 A CN 108763000A CN 201810503061 A CN201810503061 A CN 201810503061A CN 108763000 A CN108763000 A CN 108763000A
- Authority
- CN
- China
- Prior art keywords
- hard disk
- random
- delay
- requests
- asked
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/22—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing
- G06F11/2205—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing using arrangements specific to the hardware being tested
- G06F11/2221—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing using arrangements specific to the hardware being tested to test input/output devices or peripheral units
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/22—Detection or location of defective computer hardware by testing during standby operation or during idle time, e.g. start-up testing
- G06F11/26—Functional testing
- G06F11/261—Functional testing by simulating additional hardware, e.g. fault simulation
Abstract
The present invention provides a kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform, open energy saving of hard disks pattern, random I/O requests size is simulated, hard disk sector size random test is randomly assigned to different hard disks by what testing hard disk I/O was asked;The random read mode traversals of I/O, the service time difference by monitoring average latency and each I/O requests that each I/O is asked obtain the delay time domain of walker of hard disk.The present invention, in powerMode disable and enable, the pressure test of FIO is carried out according to its characteristic, judges the delay responding ability of hard disk for hard disk, effective to submit identification risk, reduces client-side issue.
Description
Technical field
The present invention relates to the technical fields of server, and in particular to a kind of hard disk I/O Request delay based on Linux platform
Simulate judgment method and system.
Background technology
Seagate is more and more stronger for the I/O requests processing capacity of high load in the application of server field, can
Stablize and efficiently the data I/O of processing high concurrent is asked.But in practical application, in addition to the demand of this high concurrent
Outside, also one kind is exactly that I/O requests are random, and the also smaller situation of I/O requests, in face of this in application, hard disk will
It is frequently converted from power save mode, active state and idle state, the conversion hard drive internal of each next state will all consume one
Fixed response time, the efficiency of this hard drive internal processing oneself state conversion can react the processing of the random I/O of hard disk
Ability.
In hard disk performance and pressure test, with the way of our generally uses be exactly to hard disk persistently transmission sequence/
The r/w of the data block of the different blockSize of machine is asked, with this come confirm hard disk sequence/it is random different
The performance and stability of r/w under blockSize.In portions of client practical application, the concurrency of the I/O requests of client is very
It is small, and the data asked are also smaller, and client has found svctm (the average service of hard disk in practical applications
Time (in milliseconds)) it can be in the presence of random the case where becoming larger.The main reason for leading to the problem of this be because hard disk from
What body Energy Saving Control mechanism (powerMode) generated, the Energy Saving Control mechanism (powerMode) of hard drive internal acquiescence is enable
State, i.e., when hard disk no I/O please after when, idle (idle) after a period of time, hard disk will be automatically into Energy Saving Control machine
It makes (powerMode), i.e., hard disk closes most functions, and the necessary function of member-retaining portion reduces the work(of hard disk with this
Consumption.When having I/O requests to enter again, hard disk will switch to active (active) state by idle (idle) state, so
Start to process data afterwards.And in this course, the energy of hard disk processing I/O requests can be reacted by two indices:
1.Svctm:The service time of average each equipment I/O request, the I/O processing capacities of disk and current load feelings are reacted
Condition, in actual test in the case of high concurrent, when load takes, svctm floats in 20-30 or so.Therefore randomness
In request, svctm reasonable values should also be within 30;2.Await:The stand-by period of average each equipment I/O operation, i.e., firmly
Disk responds delay time (tracking time+magnetic head rotation delay)+svctm service times.
Based on above-mentioned, this field judging that hard disk is delayed the technical solution of responding ability there is an urgent need for a kind of according to These parameters.
Invention content
Based on the above issues, the present invention proposes a kind of hard disk I/O Request delay simulation judgement side based on Linux platform
There is the case where IO delays by analog hard disk in method and system, the responding ability of hard disk delay are judged, accordingly according to customer demand
Identification risk can be effectively submitted, client-side issue is reduced.
The present invention provides the following technical solutions:
On the one hand, the hard disk I/O Request delay simulation judgment method based on Linux platform that the present invention provides a kind of, is wrapped
It includes:
Step 101, energy saving of hard disks pattern is opened;
Step 102, delay simulation:Simulate random I/O requests size, hard disk sector size random test;Hard disk order traversal is surveyed
Examination, hard disk random test are randomly assigned to different hard disks by what testing hard disk I/O was asked;I/O modes traverse;
Step 103, delay judgement:Random I/O is sent to ask to hard disk, interval time be more than predetermined interval after retransmit down
One I/O request, hard disk enter deep-idle state and then enter back into state of activation, average etc. by monitor that each I/O asks
Wait for the delay time domain of walker of time and the service time difference acquisition hard disk of each I/O requests.
Wherein, the I/O modes traverse as random write.
Wherein, it is described send random I/O and ask be specially to hard disk:It is automatic to execute sleep orders and waited for, hard disk into
Enter deep-idle state, cycle, which is written, in random I/O and sleep orders executes realization hard disk delay simulation, if svctm is close
Await, then I/O is almost without the stand-by period;If await is much larger than svctm, I/O queue, IOQ is long, and the response time is slack-off.
Wherein, the predetermined interval is 2min.
In addition, the present invention also provides a kind of, judgement system is simulated in the hard disk I/O Request delay based on Linux platform, described
System includes:
Control module, for opening energy saving of hard disks pattern;
Analog module, for the simulation that is delayed:Simulate random I/O requests size, hard disk sector size random test;Hard disk sequence time
Test is gone through, hard disk random test is randomly assigned to different hard disks by what testing hard disk I/O was asked;I/O modes traverse;
Judgment module is used for delay judgement:It sends random I/O to ask to hard disk, interval time is more than after predetermined interval again
Next I/O requests are sent, hard disk enters deep-idle state and then enters back into state of activation, is asked by each I/O of monitoring
The service time difference of average latency and each I/O requests obtains the delay time domain of walker of hard disk.
Wherein, the I/O modes traverse as random write.
Wherein, it is described send random I/O and ask be specially to hard disk:It is automatic to execute sleep orders and waited for, hard disk into
Enter deep-idle state, cycle, which is written, in random I/O and sleep orders executes realization hard disk delay simulation, if svctm is close
Await, then I/O is almost without the stand-by period;If await is much larger than svctm, I/O queue, IOQ is long, and the response time is slack-off.
Wherein, the predetermined interval is 2min.
The present invention provides a kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform, are opened
Energy saving of hard disks pattern, simulates random I/O requests size, and hard disk sector size random test asks testing hard disk I/O random
Distribute to different hard disks;The random read mode traversals of I/O, are asked by monitoring the average latency that each I/O is asked with each I/O
The service time difference asked obtains the delay time domain of walker of hard disk.The present invention is for hard disk in powerMode disable
When with enable, the pressure test of FIO is carried out according to its characteristic, judges the delay responding ability of hard disk, it is effective to submit identification
Risk reduces client-side issue.
Description of the drawings
Fig. 1 is flow chart of the method for the present invention;
Fig. 2 is the system structure diagram of the present invention;
Fig. 3 is the random I/O test process figure of the present invention;
Fig. 4 is the hard disk order traversal test process figure of the present invention;
Fig. 5 A-5B are the delay judgement procedure charts of the present invention.
Specific implementation mode
It to describe the technical solutions in the embodiments of the present invention more clearly, below will be to needed in the embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for ability
For the those of ordinary skill of domain, without creative efforts, it can also be obtained according to these attached drawings other attached
Figure.
The present invention a kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform, it is flat based on linux
Platform carries out powerMode disable and enable to hard disk by hard disk tool and carries out series I/O tests, and from data,
The svctm response times for finding out hard disk, to judge the delay responding ability of hard disk.
Based on above-mentioned, on the one hand, embodiments of the present invention provide a kind of hard disk I/O Request based on Linux platform and prolong
When simulate judgment method, attached drawing 1 is flow chart of the method for the present invention, including:
Step 101, energy saving of hard disks pattern is opened;
Energy saver mode opens/closes:Hard disk operational mode, is divided into energy saver mode opening, and energy saver mode closes two kinds of situations:
Energy saver mode is closed:Under such pattern, hard disk will keep active states constantly, at this point, advantage is performance highest,
It is no-delay.The disadvantage is that power consumption also higher.
Energy saver mode is opened:Under such pattern, when there is I/O requests, hard disk just enters active states.There is no I/O to ask
When asking, by taking the hard disk of Seagate as an example, hard disk acquiescence 100ms enters idle_a states (very little partial function is stopped), if
It is still asked without I/O after 2min, then hard disk enters idle_b states (partial function is stopped).When there is I/O requests, firmly
Disk then needs to wake up from the dormant state of idle_b, and idle_a wakeup times are 100ms, and idle_b wakeup times are 400ms,
Hard disk enters the completion I/O requests of active states after wake-up.Wake-up required time during this, as hard disk itself
Internal response time, and be real I/O processing times, also as delay time.From the aspect of energy saving, hard disk acquiescence is energy saving
Pattern is all opening state.At this point, advantage is energy saving.It is responded the disadvantage is that there is slight delay.
Step 102, delay simulation:Simulate random I/O requests size, hard disk sector size random test;Hard disk sequence time
Test is gone through, hard disk random test is randomly assigned to different hard disks by what testing hard disk I/O was asked;I/O modes traverse;
Since the present embodiment master is simulated into line delay, so only considering to be in the case where energy saver mode is opened when hard disk.
Parameter tune is carried out according to the feature under energy saving of hard disks pattern, in our FIO tests under normal circumstances, and simulation can be realized
Hard disk frequently occurs the phenomenon that delay.
Under normal circumstances, the main implementation feature of our hard disk FIO tests is:
A) it is tested by setting timed length
B) it is tested by setting capacity
C) BlockSize is fixed as the integer multiple of hard disk sector (512b) and is tested, such as:1k,4k,8k,16k..256k,
512k
D) hard disk order traversal is tested
E) I/O modes traverse, such as:read,randread,write,randwrite
Therefore according to the feature under energy saving of hard disks pattern, it is hard to realize that above step can be carried out following modification by we
The simulation of disk delay:
A) it is tested by setting timed length
Cancel this, because practical application scene considers that the random I/O of client must be please that the time is unfixed.
B) it is tested by setting capacity
Cancel this, because practical application scene considers that the random I/O of client must be please that capacity is unfixed.
C) BlockSize is fixed as the integer multiple of hard disk sector (512b) and is tested, such as:1k,4k,8k,16k..256k,
Fixed BlockSize is adjusted to random BlockSize by 512k, because the request of client must be random, therefore I
The case where also simulating this random I/O requests size.Implementation is to obtain the random value in range by random number,
As the size of I/O requests, hard disk is allowed to be handled, by taking 1k-1024k range random numbers as an example (as shown in Fig. 3):Upper example
For the random number in one 1-1024 of printing in 1 second interval, can see by execution twice, i.e., random BlockSize.
D) hard disk order traversal is tested
In order to which make that testing hard disk I/O asks random distributes to different hard disks so that hard disk being capable of section in different times
It is interior it is random go to respond different I/O requests, so we also use the principle of random number, make hard disk it is random appear in test
In, by taking 10 hard disks of sdb-sdk as an example, increase the random (as shown in Fig. 4) of hard disk on the basis of previous step:Upper example is 1 second
The example of output a pair of of hard disk and random BS, it can be seen that " random hard disk "+" random BS may be implemented in we
(BlockSize) " be combined with.
E) I/O modes traverse, such as:read,randread,write,randwrite
The characteristics of according to hard disk, in order to improve hard disk reading ability, hard disk is originally all the work with high speed cache, cache
It is that, when there is data write-in, data can be introduced into the high speed cache of hard disk as principle, then data will be written again by caching
Into the physical store of hard disk, during this, when hard disk high speed cache has received external data write request, i.e., can send out
Write-in is sent to complete information.Therefore, it writes and acts the improper test for going to obtain hard disk delay.Read is sequential write, hard disk tracking and
Magnetic head rotates the action for being substantially all the also sequence all on magnetic disc, therefore is also unfavorable for generating larger delay.And randread,
Tracking and magnetic head will be made to rotate and generate certain pressure, entire hard disk physical storage areas is covered, will produce in such cases
Delay situation more is changed, therefore, only retains randread herein, other I/O modes are cancelled.
To sum up, three conditions of analog hard disk delay are:BS is random, and hard disk is random, randread.
Step 103, delay judgement:Random I/O is sent to ask to hard disk, interval time be more than predetermined interval after send out again
Next I/O is sent to ask, hard disk enters deep-idle state and then enters back into state of activation, is asked by each I/O of monitoring flat
The service time difference of equal stand-by period and each I/O requests obtains the delay time domain of walker of hard disk.
The characteristics of according to hard disk node energy pattern:
In terms of feature above, idle_a sleeps for either shallow, and idle_b is deep sleep.The wakeup time of Idle is longer,
Therefore the conversion of our two states of main analog idle_b, active can be obtained the delay of 400ms or so.From script
In design, the I/O requests that we should first give hard disk one random, interval time is more than after 2min asks to next I/O again,
Enable hard disk to there is time enough to enter idle_b states, then enters back into active.In order to reach here purpose, we
Tentative I/O interval times twice are 3min.It realizes as shown in fig. 5:10 fio process have been can be seen that from attached drawing 5A
It is carrying out, the bs of each process is random.Meanwhile we can monitor the I/O states of hard disk by iostat orders, such as
Shown in attached drawing 5B:I is can be seen that out from attached drawing 5B, two disks of j do not arrive outside I/O requests, have from the other disks of b-k at random
Read-write state illustrates that other disks have come into active states.It is automatic after the completion of action above to execute the " " lives of sleep 180
It enables and waits for Wait Order to realize 180 seconds so that hard disk enters idle_b states.Then by " random I/O " above+
In " sleep180 " write-in cycle, you can realize the analogy method of hard disk delay.If svctm relatively await, explanation
I/O is almost without stand-by period (the I/O stand-by period is very short, can be ignored);If await is much larger than svctm, illustrate I/
O queues are too long, slack-off using the obtained response time.From data analysis, await that we can be monitored by iostat
It can judge hard disk with the difference (service time that the average latency-that each I/O is asked, I/O was asked every time) of svctm
What delay time float within the scope of, comparison hard disk spec, you can obtain the reasonability of hard disk delay.
The hard disk I/O Request delay simulation judgment method based on Linux platform that the present invention provides a kind of, opens hard disk section
Energy pattern, simulates random I/O requests size, and hard disk sector size random test is randomly assigned to what testing hard disk I/O was asked
Different hard disks;The random read mode traversals of I/O, by the clothes for monitoring average latency and each I/O requests that each I/O is asked
Business time difference obtains the delay time domain of walker of hard disk.The present invention for hard disk in powerMode disable and
When enable, the pressure test of FIO is carried out according to its characteristic, judges the delay responding ability of hard disk, it is effective to submit identification wind
Danger, reduces client-side issue.
On the other hand, embodiments of the present invention provide a kind of hard disk I/O Request delay simulation based on Linux platform
Judgement system, attached drawing 2 are the system structure diagrams of the present invention, the system comprises:
Control module 201, for opening energy saving of hard disks pattern;
Energy saver mode opens/closes:Hard disk operational mode, is divided into energy saver mode opening, and energy saver mode closes two kinds of situations:
Energy saver mode is closed:Under such pattern, hard disk will keep active states constantly, at this point, advantage is performance highest,
It is no-delay.The disadvantage is that power consumption also higher.
Energy saver mode is opened:Under such pattern, when there is I/O requests, hard disk just enters active states.There is no I/O to ask
When asking, by taking the hard disk of Seagate as an example, hard disk acquiescence 100ms enters idle_a states (very little partial function is stopped), if
It is still asked without I/O after 2min, then hard disk enters idle_b states (partial function is stopped).When there is I/O requests, firmly
Disk then needs to wake up from the dormant state of idle_b, and idle_a wakeup times are 100ms, and idle_b wakeup times are 400ms,
Hard disk enters the completion I/O requests of active states after wake-up.Wake-up required time during this, as hard disk itself
Internal response time, and be real I/O processing times, also as delay time.From the aspect of energy saving, hard disk acquiescence is energy saving
Pattern is all opening state.At this point, advantage is energy saving.It is responded the disadvantage is that there is slight delay.
Time delay module 202, for the simulation that is delayed:Simulate random I/O requests size, hard disk sector size random test;Firmly
Disk order traversal is tested, and hard disk random test is randomly assigned to different hard disks by what testing hard disk I/O was asked;I/O modes time
It goes through;
Since the present embodiment master is simulated into line delay, so only considering to be in the case where energy saver mode is opened when hard disk.
Parameter tune is carried out according to the feature under energy saving of hard disks pattern, in our FIO tests under normal circumstances, and simulation can be realized
Hard disk frequently occurs the phenomenon that delay.
Under normal circumstances, the main implementation feature of our hard disk FIO tests is:
A) it is tested by setting timed length
B) it is tested by setting capacity
C) BlockSize is fixed as the integer multiple of hard disk sector (512b) and is tested, such as:1k,4k,8k,16k..256k,
512k
D) hard disk order traversal is tested
E) I/O modes traverse, such as:read,randread,write,randwrite
Therefore according to the feature under energy saving of hard disks pattern, it is hard to realize that above step can be carried out following modification by we
The simulation of disk delay:
A) it is tested by setting timed length
Cancel this, because practical application scene considers that the random I/O of client must be please that the time is unfixed.
B) it is tested by setting capacity
Cancel this, because practical application scene considers that the random I/O of client must be please that capacity is unfixed.
C) BlockSize is fixed as the integer multiple of hard disk sector (512b) and is tested, such as:1k,4k,8k,16k..256k,
Fixed BlockSize is adjusted to random BlockSize by 512k, because the request of client must be random, therefore I
The case where also simulating this random I/O requests size.Implementation is to obtain the random value in range by random number,
As the size of I/O requests, hard disk is allowed to be handled, by taking 1k-1024k range random numbers as an example (as shown in Fig. 3):Upper example
For the random number in one 1-1024 of printing in 1 second interval, can see by execution twice, i.e., random BlockSize.
D) hard disk order traversal is tested
In order to which make that testing hard disk I/O asks random distributes to different hard disks so that hard disk being capable of section in different times
It is interior it is random go to respond different I/O requests, so we also use the principle of random number, make hard disk it is random appear in test
In, by taking 10 hard disks of sdb-sdk as an example, increase the random (as shown in Fig. 4) of hard disk on the basis of previous step:Upper example is 1 second
The example of output a pair of of hard disk and random BS, it can be seen that " random hard disk "+" random BS may be implemented in we
(BlockSize) " be combined with.
E) I/O modes traverse, such as:read,randread,write,randwrite
The characteristics of according to hard disk, in order to improve hard disk reading ability, hard disk is originally all the work with high speed cache, cache
It is that, when there is data write-in, data can be introduced into the high speed cache of hard disk as principle, then data will be written again by caching
Into the physical store of hard disk, during this, when hard disk high speed cache has received external data write request, i.e., can send out
Write-in is sent to complete information.Therefore, it writes and acts the improper test for going to obtain hard disk delay.Read is sequential write, hard disk tracking and
Magnetic head rotates the action for being substantially all the also sequence all on magnetic disc, therefore is also unfavorable for generating larger delay.And randread,
Tracking and magnetic head will be made to rotate and generate certain pressure, entire hard disk physical storage areas is covered, will produce in such cases
Delay situation more is changed, therefore, only retains randread herein, other I/O modes are cancelled.
To sum up, three conditions of analog hard disk delay are:BS is random, and hard disk is random, randread.
Judgment module 203 is used for delay judgement:Random I/O is sent to ask to hard disk, when interval time is more than predetermined space
Between after retransmit next I/O request, hard disk enters deep-idle state and then enters back into state of activation, passes through and monitors each I/
The service time difference of the average latency that O is asked and each I/O requests obtains the delay time domain of walker of hard disk.
The characteristics of according to hard disk node energy pattern:
In terms of feature above, idle_a sleeps for either shallow, and idle_b is deep sleep.The wakeup time of Idle is longer,
Therefore the conversion of our two states of main analog idle_b, active can be obtained the delay of 400ms or so.From script
In design, the I/O requests that we should first give hard disk one random, interval time is more than after 2min asks to next I/O again,
Enable hard disk to there is time enough to enter idle_b states, then enters back into active.In order to reach here purpose, we
Tentative I/O interval times twice are 3min.It realizes as shown in fig. 5:10 fio process have been can be seen that from attached drawing 5A
It is carrying out, the bs of each process is random.Meanwhile we can monitor the I/O states of hard disk by iostat orders, such as
Shown in attached drawing 5B:I is can be seen that out from attached drawing 5B, two disks of j do not arrive outside I/O requests, have from the other disks of b-k at random
Read-write state illustrates that other disks have come into active states.It is automatic after the completion of action above to execute the " " lives of sleep 180
It enables and waits for Wait Order to realize 180 seconds so that hard disk enters idle_b states.Then by " random I/O " above+
In " sleep180 " write-in cycle, you can realize the analogy method of hard disk delay.If svctm relatively await, explanation
I/O is almost without stand-by period (the I/O stand-by period is very short, can be ignored);If await is much larger than svctm, illustrate I/
O queues are too long, slack-off using the obtained response time.From data analysis, await that we can be monitored by iostat
It can judge hard disk with the difference (service time that the average latency-that each I/O is asked, I/O was asked every time) of svctm
What delay time float within the scope of, comparison hard disk spec, you can obtain the reasonability of hard disk delay.
The hard disk I/O Request delay simulation judgement system based on Linux platform that the present invention provides a kind of, opens hard disk section
Energy pattern, simulates random I/O requests size, and hard disk sector size random test is randomly assigned to what testing hard disk I/O was asked
Different hard disks;The random read mode traversals of I/O, by the clothes for monitoring average latency and each I/O requests that each I/O is asked
Business time difference obtains the delay time domain of walker of hard disk.The present invention for hard disk in powerMode disable and
When enable, the pressure test of FIO is carried out according to its characteristic, judges the delay responding ability of hard disk, it is effective to submit identification wind
Danger, reduces client-side issue.
The foregoing description of the disclosed embodiments enables those skilled in the art to realize or use the present invention.To this
A variety of modifications of a little embodiments will be apparent for a person skilled in the art, and the general principles defined herein can
Without departing from the spirit or scope of the present invention, to realize in other embodiments.Therefore, the present invention will not be limited
It is formed on the embodiments shown herein, but meets widest model consistent with the principles and novel features disclosed in this article
It encloses.
Claims (8)
1. a kind of hard disk I/O Request delay simulation judgment method based on Linux platform, it is characterised in that:
Step 101, energy saving of hard disks pattern is opened;
Step 102, delay simulation:Simulate random I/O requests size, hard disk sector size random test;Hard disk order traversal is surveyed
Examination, hard disk random test are randomly assigned to different hard disks by what testing hard disk I/O was asked, hard disk in different time sections with
Machine responds different I/O requests;I/O modes traverse;
Step 103, delay judgement:Random I/O is sent to ask to hard disk, interval time be more than predetermined interval after retransmit down
One I/O request, hard disk enter deep-idle state and then enter back into state of activation, average etc. by monitor that each I/O asks
Wait for the delay time domain of walker of time and the service time difference acquisition hard disk of each I/O requests.
2. according to the method described in claim 1, it is characterized in that:The I/O modes traverse as random write, tracking and magnetic head rotation
It changes the line of production raw pressure, covers entire hard disk physical storage areas.
3. according to the method described in claim 1, it is characterized in that:The random I/O of transmission is asked is specially to hard disk:Automatically
It executes sleep orders to be waited for, hard disk enters deep-idle state, and cycle, which is written, in random I/O and sleep orders executes
Realize hard disk delay simulation, if svctm is close to await, I/O is almost without the stand-by period;If await is much larger than svctm,
I/O queue, IOQ is long, and the response time is slack-off.
4. according to the method described in claim 1, it is characterized in that:The predetermined interval is 2min.
5. a kind of hard disk I/O Request delay simulation judgement system based on Linux platform, it is characterised in that:The system comprises:
Control module, for opening energy saving of hard disks pattern;
Analog module, for the simulation that is delayed:Simulate random I/O requests size, hard disk sector size random test;Hard disk sequence time
Test is gone through, hard disk random test is randomly assigned to different hard disks by what testing hard disk I/O was asked, and hard disk is in different time sections
The different I/O requests of interior random response;I/O modes traverse;
Judgment module is used for delay judgement:It sends random I/O to ask to hard disk, interval time is more than after predetermined interval again
Next I/O requests are sent, hard disk enters deep-idle state and then enters back into state of activation, is asked by each I/O of monitoring
The service time difference of average latency and each I/O requests obtains the delay time domain of walker of hard disk.
6. system according to claim 5, it is characterised in that:The I/O modes traverse as random write, tracking and magnetic head rotation
It changes the line of production raw pressure, covers entire hard disk physical storage areas.
7. system according to claim 5, it is characterised in that:The random I/O of transmission is asked is specially to hard disk:Automatically
It executes sleep orders to be waited for, hard disk enters deep-idle state, and cycle, which is written, in random I/O and sleep orders executes
Realize hard disk delay simulation, if svctm is close to await, I/O is almost without the stand-by period;If await is much larger than svctm,
I/O queue, IOQ is long, and the response time is slack-off.
8. system according to claim 5, it is characterised in that:The predetermined interval is 2min.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810503061.8A CN108763000A (en) | 2018-05-23 | 2018-05-23 | A kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810503061.8A CN108763000A (en) | 2018-05-23 | 2018-05-23 | A kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108763000A true CN108763000A (en) | 2018-11-06 |
Family
ID=64005182
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810503061.8A Pending CN108763000A (en) | 2018-05-23 | 2018-05-23 | A kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108763000A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112363901A (en) * | 2020-11-17 | 2021-02-12 | 浙江大华技术股份有限公司 | Hard disk performance detection method, device and system, storage medium and electronic device |
CN112530505A (en) * | 2020-12-29 | 2021-03-19 | 苏州元核云技术有限公司 | Hard disk delay detection method and device and computer readable storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101533366A (en) * | 2009-03-09 | 2009-09-16 | 浪潮电子信息产业股份有限公司 | Method for acquiring and analyzing performance data of server |
CN107358980A (en) * | 2017-07-19 | 2017-11-17 | 郑州云海信息技术有限公司 | A kind of method of automatic test SSD delays |
CN107799158A (en) * | 2017-11-23 | 2018-03-13 | 郑州云海信息技术有限公司 | A kind of NVMe SSD IO delays automated testing method |
CN107832202A (en) * | 2017-11-06 | 2018-03-23 | 郑州云海信息技术有限公司 | A kind of method, apparatus and computer-readable recording medium for detecting hard disk |
-
2018
- 2018-05-23 CN CN201810503061.8A patent/CN108763000A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101533366A (en) * | 2009-03-09 | 2009-09-16 | 浪潮电子信息产业股份有限公司 | Method for acquiring and analyzing performance data of server |
CN107358980A (en) * | 2017-07-19 | 2017-11-17 | 郑州云海信息技术有限公司 | A kind of method of automatic test SSD delays |
CN107832202A (en) * | 2017-11-06 | 2018-03-23 | 郑州云海信息技术有限公司 | A kind of method, apparatus and computer-readable recording medium for detecting hard disk |
CN107799158A (en) * | 2017-11-23 | 2018-03-13 | 郑州云海信息技术有限公司 | A kind of NVMe SSD IO delays automated testing method |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112363901A (en) * | 2020-11-17 | 2021-02-12 | 浙江大华技术股份有限公司 | Hard disk performance detection method, device and system, storage medium and electronic device |
CN112363901B (en) * | 2020-11-17 | 2023-08-29 | 浙江大华技术股份有限公司 | Hard disk performance detection method, device, system, storage medium and electronic device |
CN112530505A (en) * | 2020-12-29 | 2021-03-19 | 苏州元核云技术有限公司 | Hard disk delay detection method and device and computer readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20200341664A1 (en) | Intelligent wide port phy usage | |
AU2010276389B2 (en) | Adaptive power conservation in storage clusters | |
Kaushik et al. | Evaluation and analysis of greenhdfs: A self-adaptive, energy-conserving variant of the hadoop distributed file system | |
CN105205014B (en) | A kind of date storage method and device | |
Kaushik et al. | Greenhdfs: towards an energy-conserving, storage-efficient, hybrid hadoop compute cluster | |
Baek et al. | Refresh now and then | |
US7218566B1 (en) | Power management of memory via wake/sleep cycles | |
CN103270488B (en) | For the method and apparatus improving the time of restarting of platform | |
CN107003948B (en) | Electronic device and method for controlling sharable cache memory thereof | |
Zhang et al. | CREAM: A concurrent-refresh-aware DRAM memory architecture | |
CN109388339A (en) | Memory Controller and its operating method | |
US20110296098A1 (en) | System and Method for Reducing Power Consumption of Memory | |
CN108763000A (en) | A kind of hard disk I/O Request delay simulation judgment method and system based on Linux platform | |
WO2015152857A1 (en) | Energy-efficient dynamic dram cache sizing | |
US20160154452A1 (en) | System and method for controlling the power mode of operation of a memory device | |
Zhong et al. | Energy-efficient in-memory paging for smartphones | |
CN111104048B (en) | Data processing method and distributed storage system | |
US10304501B1 (en) | Implementing DRAM refresh power optimization during long idle mode | |
Hikida et al. | A power saving storage method that considers individual disk rotation | |
Xie et al. | State-dependent M/G/1/K queuing model for hard disk drives | |
US10365997B2 (en) | Optimizing DRAM memory based on read-to-write ratio of memory access latency | |
KR101803284B1 (en) | Method and apparatus for memory usage based device frequency adjustment | |
CN115826882B (en) | Storage method, device, equipment and storage medium | |
US8068373B1 (en) | Power management of memory via wake/sleep cycles | |
CN105637483A (en) | Thread migration method, apparatus and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181106 |
|
RJ01 | Rejection of invention patent application after publication |