CN104268040B - A kind of disk performance detection method and device - Google Patents
A kind of disk performance detection method and device Download PDFInfo
- Publication number
- CN104268040B CN104268040B CN201410472176.7A CN201410472176A CN104268040B CN 104268040 B CN104268040 B CN 104268040B CN 201410472176 A CN201410472176 A CN 201410472176A CN 104268040 B CN104268040 B CN 104268040B
- Authority
- CN
- China
- Prior art keywords
- disk
- feature set
- running state
- state data
- cycle
- 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.)
- Active
Links
Abstract
A kind of disk performance detection method and device, the method comprises: the running state data in n cycle before the multiple failed disk collected being broken down is divided into m group data, cycle is the time cycle that the property parameters of disk changes, m and n is positive integer; Obtain m feature set according to m group data, in m feature set, each feature set is corresponding with predetermined periodic regime; The running state data of the target disk collected is mated with m feature set respectively; If the running state data of target disk is mated with the target signature collection in m feature set, then obtain the periodic regime of answering with target signature set pair, therefore, it is possible to the residue service time of effective target of prediction disk, be convenient to process failed disk.
Description
Technical field
The present invention relates to the detection field of hardware device, particularly relate to a kind of disk performance detection method and device.
Background technology
Now, quantity of information rapidly expands, for example protects these information to need a large amount of disks.In the process that server uses, whether the proportion that hard disk failure accounts for server failure is very large, reaches more than 45%, if can break down by Accurate Prediction disk, so just can make corresponding counter-measure in advance to disk failure, be very valuable for server operation.
The Forecasting Methodology of the current disk failure had is, gather the running state data of disk, comprise the disk that can normally work and the disk damaged, wherein, the running state data of disk refers to the information of disk health and ruuning situation, and comprise disk working time, disk has turned number of times etc., and use different algorithms to set up the forecast model of disk failure, whether can break down for target of prediction disk.Therefore, in prior art, the recognition feature of feature as " being about to the dish that degenerates " of the running state data of with a large amount of disks damaged before being about to break down (as: bad before 7 days), by the running state data of a large amount of good disks recognition feature as " good coil ".When the running state data of target disk more meets the feature of " being about to the dish that degenerates ", then think that this target disk there will be fault very soon.When the running state data of target disk better meets the feature of " good dish ", will think that this disk can not break down at once.Like this, for operating disk, just can gather its running state data, and compare inside the forecast model putting into the disk failure established, and then determine whether disk can break down within a certain period of time.
But the Forecasting Methodology of existing disk failure can only predict whether disk can break down, and can not predict that disk can break down after how long, be unfavorable for the process to disk failure.
Summary of the invention
In view of this, the invention provides a kind of disk performance detection method and device, for solving the problem of the residue service time can not predicting disk of the prior art.
The disk performance detection method that the embodiment of the present invention provides, comprising:
The running state data in n cycle before the multiple failed disk collected being broken down is divided into m group data, and the described cycle is the time cycle that the property parameters of disk changes, and described m and n is positive integer;
Obtain m feature set according to described m group data, in a described m feature set, each feature set is corresponding with predetermined periodic regime;
The running state data of the target disk collected is mated with a described m feature set respectively;
If the running state data of described target disk is mated with the target signature collection in a described m feature set, then obtain the periodic regime of answering with described target signature set pair.
The disk performance pick-up unit that the embodiment of the present invention provides, comprising:
Divide module, the running state data for n the cycle before the multiple failed disk collected being broken down is divided into m group data, and the described cycle is the time cycle that the property parameters of disk changes, and described m and n is positive integer;
Processing module, after obtaining described m group data in described division module, obtain m feature set according to described m group data, in a described m feature set, each feature set is corresponding with predetermined periodic regime;
Matching module, after obtaining a described m feature set in described processing module, mates with a described m feature set respectively by the running state data of the target disk collected;
Acquisition module, if determine that the running state data of described target disk is mated with the target signature collection in a described m feature set for described matching module, then obtains the periodic regime of answering with described target signature set pair.
The disk performance detection method that the embodiment of the present invention provides, the method comprises: the running state data in n cycle before the multiple failed disk collected being broken down is divided into m group data, this cycle is the time cycle that the property parameters of disk changes, m feature set is obtained according to these m group data, in this m feature set, each feature set is corresponding with predetermined periodic regime, the running state data of the target disk collected is mated with this m feature set respectively, if the running state data of target disk is mated with the target signature collection in this m feature set, then obtain the periodic regime of answering with this target signature set pair.Mate with the running state data of target disk with periodic regime characteristic of correspondence collection by utilizing, effectively can determine the periodic regime that the target signature set pair of target disk is answered, obtain the residue service time of target disk, be convenient to the fault handling to this target disk.
For above and other object of the present invention, feature and advantage can be become apparent, preferred embodiment cited below particularly, and coordinate institute's accompanying drawings, be described in detail below.
Accompanying drawing explanation
Fig. 1 a is the schematic diagram of the structure of server in the embodiment of the present invention;
Fig. 1 b is the schematic diagram of the structure of personal computer in the embodiment of the present invention;
Fig. 2 is a schematic diagram of disk performance detection method in the embodiment of the present invention;
Fig. 3 is another schematic diagram of disk performance detection method in the embodiment of the present invention;
Fig. 4 is a schematic diagram of the division of running state data in this numb embodiment;
Fig. 5 is a schematic diagram of disk performance pick-up unit in the embodiment of the present invention;
Fig. 6 is another schematic diagram of disk performance pick-up unit in the embodiment of the present invention.
Embodiment
For further setting forth the present invention for the technological means that realizes predetermined goal of the invention and take and effect, below in conjunction with accompanying drawing and preferred embodiment, to according to the specific embodiment of the present invention, structure, feature and effect thereof, be described in detail as follows.
In embodiments of the present invention, disk performance pick-up unit can be the server of hyperdisk, refer to 1a, for in the embodiment of the present invention, the schematic diagram of the structure of server, this server 100 can produce larger difference because of configuration or performance difference, one or more central processing units (centralprocessingunits can be comprised, CPU) 122 (such as, one or more processors) and storer 132, one or more store the storage medium 130 (such as one or more mass memory units) of application program 142 or data 144.Wherein, storer 132 and storage medium 130 can be of short duration storages or store lastingly.The program being stored in storage medium 130 can comprise one or more modules (illustrating not shown), and each module can comprise a series of command operatings in server.Further, central processing unit 122 can be set to communicate with storage medium 130, and server 100 performs a series of command operatings in storage medium 130.Server 100 can also comprise one or more power supplys 126, one or more wired or wireless network interfaces 150, one or more IO interface 158, and/or, one or more operating system 141, such as WindowsServerTM, MacOSXTM, UnixTM, LinuxTM, FreeBSDTM etc.Step in the embodiment of the present invention performed by disk performance pick-up unit can based on the server architecture shown in Fig. 1 a.
It should be noted that, in embodiments of the present invention, disk performance pick-up unit can also be personal computer, refer to 1b, for in the embodiment of the present invention, the schematic diagram of the structure of personal computer, as shown in Figure 1 b, personal computer 200 comprises storer 102, processor 104, memory controller 106, Peripheral Interface 108, mixed-media network modules mixed-media 110, display module 112 and sensor 114.Be appreciated that the structure shown in Fig. 1 b is only signal, it does not cause restriction to the structure of personal computer 200.Such as, personal computer 200 also can comprise than assembly more or less shown in Fig. 1 b, or has the configuration different from shown in Fig. 1 b.
Storer 102 can be used for storing software program and module, as the disk performance detection method in the embodiment of the present invention and programmed instruction/module corresponding to device, processor 104 is by running the software program and module that are stored in storer 102, thus perform the application of various function and data processing, namely realize above-mentioned method.
Storer 102 can comprise high speed random access memory, also can comprise nonvolatile memory, as one or more magnetic storage device, flash memory or other non-volatile solid state memories.In some instances, storer 102 can comprise the storer relative to the long-range setting of processor 106 further, and these remote memories can be connected to personal computer 200 by network.The example of above-mentioned network includes but not limited to internet, intranet, LAN (Local Area Network), mobile radio communication and combination thereof.Processor 106 and other possible assemblies can carry out the access of storer 102 under the control of memory controller 104.
Various input/output device is coupled to processor 106 by Peripheral Interface 108.Various softwares in processor 106 run memory 102, instruction personal computer 200 perform various function and carry out data processing.In certain embodiments, Peripheral Interface 108, processor 106 and memory controller 104 can realize in one single chip.In some other example, they can respectively by independently chip realization.
Mixed-media network modules mixed-media 110 is for receiving and sending network signal.Above-mentioned network signal can comprise wireless signal or wire signal.In an example, above-mentioned network signal is cable network signal.Now, mixed-media network modules mixed-media 110 can comprise the elements such as processor, random access memory, converter, crystal oscillator.In one embodiment, above-mentioned network signal is wireless signal (such as radiofrequency signal).Now mixed-media network modules mixed-media 110 essence is radio-frequency module, receives and sends electromagnetic wave, realizing the mutual conversion of electromagnetic wave and electric signal, thus carry out communication with communication network or other equipment.Radio-frequency module can comprise the various existing circuit component for performing these functions, such as, and antenna, radio-frequency (RF) transceiver, digital signal processor, encrypt/decrypt chip, subscriber identity module (SIM) card, storer etc.Radio-frequency module can with various network as internet, intranet, wireless network carry out communication or carry out communication by wireless network and other equipment.Above-mentioned wireless network can comprise cellular telephone networks, WLAN (wireless local area network) or Metropolitan Area Network (MAN).Above-mentioned wireless network can use various communication standard, agreement and technology, include, but are not limited to global system for mobile communications (GlobalSystemforMobileCommunication, GSM), enhancement mode mobile communication technology (EnhancedDataGSMEnvironment, EDGE), Wideband CDMA Technology (widebandcodedivisionmultipleaccess, W-CDMA), CDMA (Code Division Multiple Access) (Codedivisionaccess, CDMA), tdma (timedivisionmultipleaccess, TDMA), adopting wireless fidelity technology (Wireless, Fidelity, WiFi) (as IEEE-USA standard IEEE 802.11a, IEEE802.11b, IEEE802.11g and/or IEEE802.11n), the networking telephone (Voiceoverinternetprotocal, VoIP), worldwide interoperability for microwave access (WorldwideInteroperabilityforMicrowaveAccess, Wi-Max), other are for mail, the agreement of instant messaging and short message, and any other suitable communications protocol, even can comprise those current agreements be developed not yet.
Display module 112 is for showing the various graphical user interface of the information inputted by user, the information being supplied to user and personal computer 200, and these graphical user interface can be made up of figure, text, icon, video and its combination in any.In an example, display module 112 comprises a display panel.Display panel such as can be a display panels (LiquidCrystalDisplay, LCD), Organic Light Emitting Diode (OrganicLight-EmittingDiodeDisplay, OLED) display panel, electrophoretic display panel (Electro-PhoreticDisplay, EPD) etc.Further, touch-control surface can be arranged on display panel thus to form an entirety with display panel.In further embodiments, display module 112 also can comprise the display device of other types, such as, comprise a projection display equipment.Compared to general display panel, projection display equipment also needs to comprise some parts such as lens combination for projecting.
The example of sensor 114 includes, but are not limited to: camera, iris capturing device, fingerprint capturer, microphone.Camera is used for taking pictures or video, and camera specifically can comprise the assemblies such as camera lens module, image sensor and flashlamp.Camera lens module is used for the target imaging be taken, and imaging is mapped in image sensor.Image sensor, for receiving the light from camera lens module, realizes photosensitive, with recording image information.Particularly, image sensor can based on complementary metal oxide semiconductor (CMOS) (ComplementaryMetalOxideSemiconductor, CMOS), charge coupled cell (Charge-coupledDevice, CCD) or other image sensing principles realize.Flashlamp is used for carrying out exposure compensating when taking.In general, the flashlamp for personal computer 200 can be light-emittingdiode (LightEmittingDiode, LED) flashlamp.
Iris capturing device is for gathering the iris of user, and it for the parts set up separately, also by uniting two into one with camera, that is can use as iris capturing device while of camera.
Fingerprint adopts device for gathering the fingerprint of user, and it can be the parts set up separately, can also be to be integrated in other parts.Such as, in one embodiment, the display panel of display module 112 is integrated with image sensor simultaneously, can sense the image of the object on display module 112 surface, and now, display module 112 can adopt device to use as fingerprint simultaneously.
Refer to Fig. 2, be the embodiment of disk performance detection method in the embodiment of the present invention, comprise:
201, the multiple failed disk collected are broken down before the running state data in n cycle be divided into m group data, the cycle is time cycle of the property parameters change of disk, m and n is positive integer;
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by disk performance pick-up unit (hereinafter referred to as pick-up unit) is divided into m group data, wherein, cycle is the time cycle that the property parameters of disk changes, m and n is positive integer.
Wherein, running state data can be the SMART data of disk, and the disk hardware information that during SMART data, disk self-checking system retains, comprises the data of hard disk health status and ruuning situation, and such as disk working time, disk run up number of times etc.
Wherein, the property parameters of disk refers to have periodically variable property parameters, and the property parameters of disk can be: bottom data read error rate (readerrorrate), seek error rate (SeekErrorRate), the sector count (UncorrectableSectorCount) that cannot correct or senior direct memory access cyclic redundancy check (CRC) code error count (UltraDMACRCErrorCount).
In embodiments of the present invention, prediction unit can calculate the cycle of the periodically variable parameter of this disk according to the running state data of the disk collected, such as bottom data read error rate, when adding up bottom data read error rate, only have and just can calculate bottom data read error rate when the data read reach predetermined value, and again calculate, therefore, the bottom data read error rate of disk is relevant with the read-write handling capacity of disk, and the period of change of bottom data read error rate is less, show that change is faster, read-write speed is larger.
202, obtain m feature set according to m group data, in m feature set, each feature set is corresponding with predetermined periodic regime;
In embodiments of the present invention, prediction unit will obtain m feature set according to m group data, and in this m feature set, each feature set is corresponding with predetermined periodic regime, and wherein, periodic regime is relevant with n cycle.
203, the running state data of the target disk collected is mated with m feature set respectively;
If the running state data of 204 target disks is mated with the target signature collection in m feature set, then obtain the periodic regime of answering with target signature set pair.
In embodiments of the present invention, prediction unit is after obtaining m feature set, the running state data of the target disk collected is mated with this m feature set respectively, if and the running state data of target disk is mated with the target signature collection in this m feature set, then obtain the periodic regime of answering with this target signature set pair.
Wherein, target signature integrates as feature set the highest with the running state data matching degree of target disk in m feature set.
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by prediction unit is divided into m group data, this cycle is the time cycle that the property parameters of disk changes, m feature set is obtained according to these m group data, in this m feature set, each feature set is corresponding with predetermined periodic regime, the running state data of the target disk collected is mated with this m feature set respectively, if the running state data of target disk is mated with the target signature collection in this m feature set, then obtain the periodic regime of answering with this target signature set pair.Mate with the running state data of target disk with periodic regime characteristic of correspondence collection by utilizing, effectively can determine the periodic regime that the target signature set pair of target disk is answered, obtain the residue service time of target disk, be convenient to the fault handling to this target disk.
Technical scheme for a better understanding of the present invention in embodiment, refers to Fig. 3, is the embodiment of disk performance detection method in the embodiment of the present invention, comprises:
301, the multiple failed disk collected are broken down before the running state data in n cycle be on average divided into m group data according to the cycle, comprise the identical of multiple failed disk and the running state data in a continuous print n/m cycle in each group data, n/m is positive integer;
In embodiments of the present invention, prediction unit target of prediction disk whether be about to become failed disk and residue service time of this target disk time, need the running state data utilizing failed disk before breaking down to set up feature set.The running state data in n cycle before the multiple failed disk collected break down by prediction unit is on average divided into m group data according to the cycle, wherein, comprise the identical of multiple failed disk and the running state data in a continuous print n/m cycle in each group data, n/m is positive integer.Such as: if prediction unit collection 100 failed disk break down before the running state data in 25 cycles, and be on average divided into 5 groups of data according to the cycle, concrete, the running state data in the 1st to the 5th cycle of these 100 failed disk is first group of data, the running state data in the 6th to the 10th cycle of these 100 failed disk is second group of data, the running state data in the 11st to the 15th cycle of these 100 failed disk is the 3rd group of data, the running state data in the 16th to the 20th cycle of these 100 failed disk is the 4th group of data, the running state data in the 21 to 25 cycle of these 100 failed disk is the 5th group of data.Wherein, the numbering in the cycle in the embodiment of the present invention calculates from the time point broken down, such as: the 1st refers to the one-period duration before breaking down, in the 5th cycle, is distance and breaks down the duration in 5 cycles in addition.
In embodiments of the present invention, m group data are divided into according to the cycle by the running state data in n cycle before multiple failed disk being broken down, make it possible to the feature of the running state data obtained at different periodic regimes, and can be used for target of prediction disk residue service time.
It should be noted that, in embodiments of the present invention, the running state data on average dividing n cycle according to the cycle is only a kind of feasible implementation, prediction unit can also according to the rule pre-set the running state data in non-average this n of division cycle.
302, support vector machine many sorting techniques process m group data are utilized to obtain m feature set; And the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect obtains the feature set of normal disk;
In embodiments of the present invention, prediction unit obtains m feature set by utilizing these m group data of support vector machine many sorting techniques process, and the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect obtains the feature set of normal disk.Each feature set wherein in m feature set is corresponding with periodic regime, and the periodic regime of this periodic regime one group of data that to be this feature set corresponding, such as: if first group of packet contains: the running state data in 1st to 5th cycle of multiple failed disk before breaking down, then the periodic regime obtaining the feature set of these first group of data corresponding is the 1st to the 5th cycle.
303, the running state data of the target disk collected is mated with the feature set of m feature set and normal disk respectively; Perform step 304 and 306 respectively;
In embodiments of the present invention, after the feature set of m the feature set and normal disk that obtain failed disk, the running state data of the target disk collected is mated with the feature set of this m feature set and normal disk by prediction unit respectively, be specially: prediction unit utilizes the running state data of this target disk of support vector machine many sorting techniques process, obtain the feature of this target disk, the feature of this target disk is mated with the feature in the feature set of m feature set and normal disk respectively.
If the running state data of 304 target disks is mated with the target signature collection in m feature set, then obtain the periodic regime of answering with target signature set pair;
305, using the residue service time of the product of the periodic regime of the duration of the one-period of target disk and target signature collection as target disk;
In embodiments of the present invention, if target signature integrates as in m feature set, then obtain the periodic regime of answering with this target signature set pair, and calculate the residue service time of target disk according to the duration of the one-period of target disk and the periodic regime of target signature collection.
Wherein, target signature collection is the feature set the highest with the matching degree of the running state data of target disk.And the duration of the one-period of target disk is the cycle of carrying out the collection of running state data with failed disk is identical property parameters, such as, if the running status operation parameter in n cycle of the failed disk gathered uses the cycle of bottom data read error rate, then the one-period of this target disk is also the cycle of the bottom data read error rate of this target disk.
Wherein, the mode calculating residue service time of target disk is exemplified as: if periodic regime be K cycle to L cycle, and the duration in each cycle is T, then the residue of this target disk is service time: (K-1) T to LT.
Such as: the periodic regime that target signature set pair is answered is the 5th to the 10th cycle, and this cycle is the cycle of bottom data read error rate, and the cycle of target disk bottom data read error rate is 2 hours, then residue service time of this target disk is 8 to 20 hours, and namely 8 to 20 hours in future break down by this target disk.
If the running state data of 306 target disks is mated with the feature set of normal disk, then determine that target disk is normal disk.
In embodiments of the present invention, if the running state data of target disk is mated with the feature set of normal disk, then determine that target disk is normal disk.
Wherein, mate the matching degree of the feature set of the normal disk of data field of the operation referring to target disk with the feature set of normal disk the highest for the running state data of target disk.
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by prediction unit is on average divided into m group data according to the cycle, the identical of multiple failed disk is all comprised and the running state data in a continuous print n/m cycle in each group data, and utilize these m group data of support vector machine many sorting techniques process to obtain m corresponding feature set, and the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect obtains the feature set of normal disk, the running state data of the target disk obtained gathered is mated with the feature set of this m feature set and normal disk respectively, if the feature set the highest with the running state data matching degree of this target disk and target signature collection are one in this m feature set, then obtain the periodic regime that this target signature set pair is answered, using the residue service time of the product of the periodic regime of the duration of the one-period of target disk and this target signature collection as target disk, therefore, effectively can obtain the time that this target disk distance breaks down, be convenient to process the fault of this target disk.
Technical scheme for a better understanding of the present invention in embodiment, will a concrete application scenarios be introduced below:
The running state data in 25 cycles before collect 4 disk failures is on average divided into 5 groups of data according to the cycle by prediction unit, this cycle refers to the period of change of the bottom data read error rate of disk, refer to Fig. 4, for under this application scene by the schematic diagram of running state data division group, 4 disks are comprised in Fig. 4, each horizontal line represents the time shaft of a disk failures, the bad time point representing disk failures, the running state data of disk from the cycle of the 26th before breaking down is worked as the running state data carrying out dish by good expression, as shown in FIG., using the running state data of the service data that the cycle starts of the 26th before disk failures as normal disk, be the 1st group of data, the running state data in the before failed disk being broken down the 21st to the 25th cycle is divided into the 2nd group of data, the running state data in the 16th to the 20th cycle is divided into the 3rd group of data, the running state data in the 11st to the 15th cycle is divided into the 4th group of data, the running state data in the 6th to the 10th cycle is divided into the 5th group of data, by the 1st to the 5th cycle running state data be divided into the 6th group of data.
Utilize these 6 groups of data of support vector machine many sorting techniques process, obtain the feature set of these 6 groups of data respectively, the running state data of the target disk collected is mated with the feature set of these 6 groups of data respectively, if the matching degree of the feature set of the running state data of target disk and the 1st group of data is the highest, then this target disk is normal disk.
With the cycle of the bottom data read error rate of target disk for 3 hours, if then the matching degree of the running state data of target disk and the feature set of the 2nd group of data is the highest, then this target disk distance breaks down 21 to 25 cycles in addition, remaining service time is 60 to 75 hours, if the matching degree of the feature set of the running state data of target disk and the 3rd group of data is the highest, then this target disk distance breaks down 16 to 20 cycles in addition, remaining service time is 45 to 60 hours, if the matching degree of the feature set of the running state data of target disk and the 4th group of data is the highest, then this target disk distance breaks down 11 to 15 cycles in addition, then remaining service time is 30 to 45 hours, if the matching degree of the feature set of the running state data of this target disk and the 5th group of data is the highest, then this target disk distance breaks down 6 to 10 cycles in addition, then remain 15 to 30 hours service time, if for the matching degree of the running state data of this target disk and the feature set of the 6th group of data is the highest, then this target disk distance breaks down 1 to 5 cycle in addition, then remaining service time is 0 to 15 hour.
Refer to Fig. 5, be a kind of in embodiment of the present invention embodiment of structure of disk performance pick-up unit, comprise:
Divide module 501, the running state data for n the cycle before the multiple failed disk collected being broken down is divided into m group data, and the cycle is the time cycle that the property parameters of disk changes, m and n is positive integer;
Processing module 502, after obtaining m group data in division module, obtain m feature set according to m group data, in m feature set, each feature set is corresponding with predetermined periodic regime;
Matching module 503, for obtain m feature set in processing module after, mates with m feature set respectively by the running state data of the target disk collected;
Acquisition module 504, if mate with the target signature collection in m feature set for the running state data of matching module determination target disk, then obtains the periodic regime of answering with target signature set pair.
Wherein, running state data can be the SMART data of disk, and the disk hardware information that during SMART data, disk self-checking system retains, comprises the data of hard disk health status and ruuning situation, and such as disk working time, disk run up number of times etc.
Wherein, target signature integrates as feature set the highest with the running state data matching degree of target disk in m feature set.
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by the division module 501 in prediction unit is divided into m group data, cycle is the time cycle that the property parameters of disk changes, m and n is positive integer; Then, processing module 502 obtains m feature set according to m group data, and in m feature set, each feature set is corresponding with predetermined periodic regime; And by matching module 503, the running state data of the target disk collected is mated with m feature set respectively; If matching module 503 determines that the running state data of target disk is mated with the target signature collection in m feature set, then acquisition module 504 obtains the periodic regime of answering with target signature set pair.
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by prediction unit is divided into m group data, this cycle is the time cycle that the property parameters of disk changes, m feature set is obtained according to these m group data, in this m feature set, each feature set is corresponding with predetermined periodic regime, the running state data of the target disk collected is mated with this m feature set respectively, if the running state data of target disk is mated with the target signature collection in this m feature set, then obtain the periodic regime of answering with this target signature set pair.Mate with the running state data of target disk with periodic regime characteristic of correspondence collection by utilizing, effectively can determine the periodic regime that the target signature set pair of target disk is answered, obtain the residue service time of target disk, be convenient to the fault handling to this target disk.
Prediction unit for a better understanding of the present invention in embodiment, refer to Fig. 6, for another embodiment of the structure of prediction unit in the embodiment of the present invention, comprise: the division module 501 as shown in Figure 5 in embodiment, processing module 502, matching module 503, acquisition module 504, and similar to the content of middle description embodiment illustrated in fig. 5, do not repeat herein.
In embodiments of the present invention, divide module 501 specifically for:
The running state data in n cycle before the multiple failed disk collected being broken down on average is divided into m group data according to the cycle, comprise the identical of multiple failed disk and the running state data in a continuous print n/m cycle in each group data, n/m is positive integer.
In embodiments of the present invention, prediction unit also comprises:
Computing module 601, after obtaining the periodic regime of target signature collection at acquisition module 504, calculates the residue service time of target disk according to the duration of the one-period of target disk and the periodic regime of target signature collection.
In embodiments of the present invention, processing module 502 obtains m feature set specifically for utilizing support vector machine many sorting techniques process m group data.
And processing module 502 also obtains the feature set of normal disk for the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect;
Matching module 503 is also for mating the feature set of the running state data of target disk with normal disk;
And in embodiments of the present invention, prediction unit also comprises:
Determination module 602, if determine that the running state data of target disk is mated with the feature set of normal disk for matching module 503, then determines that target disk is normal disk.
Wherein, running state data can be the SMART data of disk, and the disk hardware information that during SMART data, disk self-checking system retains, comprises the data of hard disk health status and ruuning situation, and such as disk working time, disk run up number of times etc.
In embodiments of the present invention, the property parameters of disk is: bottom data read error rate, seek error rate, the sector count that cannot correct or senior direct memory access cyclic redundancy check (CRC) code error count.
It should be noted that, in embodiments of the present invention, the running state data on average dividing n cycle according to the cycle is only a kind of feasible implementation, prediction unit can also according to the rule pre-set the running state data in non-average this n of division cycle.
Wherein, target signature collection is the feature set the highest with the matching degree of the running state data of target disk.And the duration of the one-period of target disk is the cycle of carrying out the collection of running state data with failed disk is identical property parameters, such as, if the running status operation parameter in n cycle of the failed disk gathered uses the cycle of bottom data read error rate, then the one-period of this target disk is also the cycle of the bottom data read error rate of this target disk.
Wherein, the mode calculating residue service time of target disk is exemplified as: if periodic regime be K cycle to L cycle, and the duration in each cycle is T, then the residue of this target disk is service time: (K-1) T to LT.
Such as: the periodic regime that target signature set pair is answered is the 5th to the 10th cycle, and this cycle is the cycle of bottom data read error rate, and the cycle of target disk bottom data read error rate is 2 hours, then residue service time of this target disk is 8 to 20 hours, and namely 8 to 20 hours in future break down by this target disk.
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by the division module 501 in prediction unit is on average divided into m group data according to the cycle, comprise the identical of multiple failed disk and the running state data in a continuous print n/m cycle in each group data, n/m is positive integer; Then, processing module 502 utilizes support vector machine many sorting techniques process m group data to obtain m feature set, and the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect obtains the feature set of normal disk, wherein in m feature set, each feature set is corresponding with predetermined periodic regime; And by matching module 503, the running state data of the target disk collected is mated with the feature set of m feature set and normal disk respectively; If matching module 503 determines that the running state data of target disk is mated with the target signature collection in m feature set, then acquisition module 504 obtains the periodic regime of answering with target signature set pair, and calculates the residue service time of target disk according to the periodic regime of the duration of the one-period of target disk and target signature collection by computing module 601.If matching module 503 determines that the running state data of target disk is mated with the feature set of normal disk, then determination module 602 determines that target disk is normal disk.
In embodiments of the present invention, the running state data in n cycle before the multiple failed disk collected break down by prediction unit is on average divided into m group data according to the cycle, the identical of multiple failed disk is all comprised and the running state data in a continuous print n/m cycle in each group data, and utilize these m group data of support vector machine many sorting techniques process to obtain m corresponding feature set, and the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect obtains the feature set of normal disk, the running state data of the target disk obtained gathered is mated with the feature set of this m feature set and normal disk respectively, if the feature set the highest with the running state data matching degree of this target disk and target signature collection are one in this m feature set, then obtain the periodic regime that this target signature set pair is answered, using the residue service time of the product of the periodic regime of the duration of the one-period of target disk and this target signature collection as target disk, therefore, effectively can obtain the time that this target disk distance breaks down, be convenient to process the fault of this target disk.
The above, it is only preferred embodiment of the present invention, not any pro forma restriction is done to the present invention, although the present invention discloses as above with preferred embodiment, but and be not used to limit the present invention, any those skilled in the art, do not departing within the scope of technical solution of the present invention, make a little change when the technology contents of above-mentioned announcement can be utilized or be modified to the Equivalent embodiments of equivalent variations, in every case be do not depart from technical solution of the present invention content, according to any simple modification that technical spirit of the present invention is done above embodiment, equivalent variations and modification, all still belong in the scope of technical solution of the present invention.
Claims (12)
1. a disk performance detection method, is characterized in that, comprising:
The running state data in n cycle before the multiple failed disk collected being broken down is divided into m group data, and the described cycle is the time cycle that the property parameters of disk changes, and described m and n is positive integer;
Obtain m feature set according to described m group data, in a described m feature set, each feature set is corresponding with predetermined periodic regime;
The running state data of the target disk collected is mated with a described m feature set respectively;
If the running state data of described target disk is mated with the target signature collection in a described m feature set, then obtain the periodic regime of answering with described target signature set pair.
2. method according to claim 1, is characterized in that, described the multiple failed disk collected are broken down before the running state data in n cycle be divided into m group data and comprise:
The running state data in n cycle before the multiple failed disk collected being broken down on average is divided into m group data according to the cycle, comprise the identical of described multiple failed disk and the running state data in a continuous print n/m cycle in each group data, described n/m is positive integer.
3. method according to claim 1, is characterized in that, described method also comprises:
After obtaining the periodic regime of answering with described target signature set pair, calculate the residue service time of described target disk according to the duration of the one-period of described target disk and the periodic regime of described target signature collection.
4. the method according to claims 1 to 3 any one, is characterized in that, describedly obtains m feature set according to described m group data and comprises:
M group data described in the process of support vector machine many sorting techniques are utilized to obtain m feature set.
5. method according to claim 4, is characterized in that, described method also comprises:
Before the running state data of the target disk collected being mated with a described m feature set respectively, the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect obtains the feature set of described normal disk;
When the running state data of the target disk collected is mated with a described m feature set respectively, also further the feature set of the running state data of described target disk with described normal disk is mated;
If the running state data of described target disk is mated with the feature set of described normal disk, then determine that described target disk is normal disk.
6. method according to claim 1, is characterized in that, the property parameters of described disk is: bottom data read error rate, seek error rate, the sector count that cannot correct or senior direct memory access cyclic redundancy check (CRC) code error count.
7. a disk performance pick-up unit, is characterized in that, comprising:
Divide module, the running state data for n the cycle before the multiple failed disk collected being broken down is divided into m group data, and the described cycle is the time cycle that the property parameters of disk changes, and described m and n is positive integer;
Processing module, after obtaining described m group data in described division module, obtain m feature set according to described m group data, in a described m feature set, each feature set is corresponding with predetermined periodic regime;
Matching module, after obtaining a described m feature set in described processing module, mates with a described m feature set respectively by the running state data of the target disk collected;
Acquisition module, if determine that the running state data of described target disk is mated with the target signature collection in a described m feature set for described matching module, then obtains the periodic regime of answering with described target signature set pair.
8. device according to claim 7, is characterized in that, divide module specifically for:
The running state data in n cycle before the multiple failed disk collected being broken down on average is divided into m group data according to the cycle, comprise the identical of described multiple failed disk and the running state data in a continuous print n/m cycle in each group data, described n/m is positive integer.
9. device according to claim 7, is characterized in that, described device also comprises:
Computing module, for obtain described target signature collection at described acquisition module periodic regime after, calculate the residue service time of described target disk according to the duration of the one-period of described target disk and the periodic regime of described target signature collection.
10. the device according to claim 7 to 9 any one, is characterized in that, described processing module obtains m feature set specifically for utilizing m group data described in the process of support vector machine many sorting techniques.
11. devices according to claim 10, is characterized in that, described processing module also obtains the feature set of described normal disk for the running state data of the multiple normal disk utilizing the process of support vector machine many sorting techniques to collect;
Described matching module is also for mating the feature set of the running state data of described target disk with described normal disk;
Described device also comprises:
Determination module, if determine that the running state data of described target disk is mated with the feature set of described normal disk for described matching module, then determines that described target disk is normal disk.
12. devices according to claim 7, is characterized in that, the property parameters of described disk is: bottom data read error rate, seek error rate, the sector count that cannot correct or senior direct memory access cyclic redundancy check (CRC) code error count.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410472176.7A CN104268040B (en) | 2014-09-16 | 2014-09-16 | A kind of disk performance detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410472176.7A CN104268040B (en) | 2014-09-16 | 2014-09-16 | A kind of disk performance detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104268040A CN104268040A (en) | 2015-01-07 |
| CN104268040B true CN104268040B (en) | 2016-04-06 |
Family
ID=52159563
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410472176.7A Active CN104268040B (en) | 2014-09-16 | 2014-09-16 | A kind of disk performance detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104268040B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105068901A (en) * | 2015-07-27 | 2015-11-18 | 浪潮电子信息产业股份有限公司 | Disk detection method |
| CN107025153B (en) * | 2016-01-29 | 2021-02-12 | 阿里巴巴集团控股有限公司 | Disk failure prediction method and device |
| CN109976971B (en) * | 2017-12-28 | 2023-08-04 | 北京京东尚科信息技术有限公司 | Hard disk state monitoring method and device |
| CN108205424B (en) * | 2017-12-29 | 2021-09-28 | 北京奇虎科技有限公司 | Data migration method and device based on disk and electronic equipment |
| CN108681496A (en) * | 2018-05-09 | 2018-10-19 | 北京奇艺世纪科技有限公司 | Prediction technique, device and the electronic equipment of disk failure |
| CN110164501B (en) * | 2018-06-29 | 2022-05-20 | 腾讯科技（深圳）有限公司 | Hard disk detection method, device, storage medium and equipment |
| CN109496292B (en) * | 2018-10-16 | 2022-02-22 | 深圳市锐明技术股份有限公司 | Disk management method, disk management device and electronic equipment |
| CN112527572A (en) * | 2019-09-19 | 2021-03-19 | 中兴通讯股份有限公司 | Disk failure prediction method and device, computer readable storage medium and server |
| CN112463486A (en) * | 2020-11-24 | 2021-03-09 | 苏州浪潮智能科技有限公司 | Disk quality inspection method, system, terminal and storage medium |
| CN115509465B (en) * | 2022-11-21 | 2023-03-28 | 杭州字节方舟科技有限公司 | Sector management method and device, electronic equipment and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103761180A (en) * | 2014-01-11 | 2014-04-30 | 浪潮电子信息产业股份有限公司 | Method for preventing and detecting disk faults during cluster storage |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070153413A1 (en) * | 2005-12-30 | 2007-07-05 | Hiromichi Kuwano | Active head fault detection scheme |
| JP4962116B2 (en) * | 2007-04-25 | 2012-06-27 | 船井電機株式会社 | Disk unit |
-
2014
- 2014-09-16 CN CN201410472176.7A patent/CN104268040B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103761180A (en) * | 2014-01-11 | 2014-04-30 | 浪潮电子信息产业股份有限公司 | Method for preventing and detecting disk faults during cluster storage |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104268040A (en) | 2015-01-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104268040B (en) | A kind of disk performance detection method and device | |
| US10915156B2 (en) | Wearable device-aware supervised power management for mobile platforms | |
| US11423695B2 (en) | Face location tracking method, apparatus, and electronic device | |
| US11095689B2 (en) | Service processing method and apparatus | |
| CN108242837B (en) | Electronic device and method of controlling charging of electronic device | |
| US20160210632A1 (en) | Secured payment method and relevant device and system | |
| TWI503025B (en) | Cloud-aware collaborative mobile platform power management using mobile sensors | |
| CN104618217A (en) | Method, terminal, server and system for sharing resource | |
| CN104954631A (en) | Video processing method, device and system | |
| CN104867095A (en) | Image processing method and device | |
| CN104639498A (en) | Fingerprint matching method, device and system | |
| CN103177217B (en) | A kind of file scanning method, system and client and server | |
| US20170048481A1 (en) | Electronic device and image encoding method of electronic device | |
| CN105187710A (en) | Photography controlling method, intelligent watch, photography terminal and system | |
| US10491884B2 (en) | Image processing method and electronic device supporting the same | |
| CN107832142A (en) | The resource allocation method and equipment of application program | |
| EP3371994A1 (en) | Apparatus, method and computer program product for privacy protection | |
| CN105120158A (en) | Shooting method and apparatus for mobile terminal | |
| CN112954212B (en) | Video generation method, device and equipment | |
| US10212291B2 (en) | System, method, and non-transitory computer readable storage medium for image recognition based on convolutional neural networks | |
| CN106210514A (en) | Take pictures focusing method, device and smart machine | |
| EP3461138A1 (en) | Processing method and terminal | |
| CN103618625A (en) | Method, device and system for controlling network resources in microblog | |
| CN109992369A (en) | Applied program processing method and device, electronic equipment, computer readable storage medium | |
| CN110751003B (en) | Method and equipment for acquiring target data information of two-dimension code |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |