CN105320461A - Self-adaption fast reaction control system for software definition storage system - Google Patents
Self-adaption fast reaction control system for software definition storage system Download PDFInfo
- Publication number
- CN105320461A CN105320461A CN201410312060.7A CN201410312060A CN105320461A CN 105320461 A CN105320461 A CN 105320461A CN 201410312060 A CN201410312060 A CN 201410312060A CN 105320461 A CN105320461 A CN 105320461A
- Authority
- CN
- China
- Prior art keywords
- self
- adaptation
- module
- response control
- value
- 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.)
- Granted
Links
Abstract
The invention discloses a self-adaption fast reaction control system for a software definition storage system. The self-adaption fast reaction control system is used for improving performance parameters. The system comprises a flow monitoring module, a self-adaption double-nerve module and a fast response control module. The flow monitoring module is used for acquiring an observed value of the performance parameters on a storage node. The self-adaption double-nerve module is used for learning the optimum allocation of a plurality of storage devices in the storage node from the historical records and related observed values of allocations of the storage device between the observed value and the specific values of the performance parameters under the condition of different difference values, and when an existing difference value is not smaller than a threshold value, the optimum allocation is provided. If the existing difference value is not smaller than the threshold value, the fast response control module is used for changing the optimum allocation of the storage devices provided by the self-adaption double-nerve module and adopted as the existing allocation of the storage devices in the storage node.
Description
Technical field
The present invention, about a kind of control system stored for software definition, is particularly used for software definition about one and stores, to reach the control system of the particular characteristic pointer of demand in service level agreement.
Background technology
Cloud service develops very universal in last decade.Cloud service calculates based on high in the clouds, under the burden situation not increasing client, provides relevant service or commodity.High in the clouds calculates and relate to a large amount of main frames, and these main frames each other via telecommunication network, such as a Internet, and connect.It relies on sharing, with compliance and economic scale of resource.The concept that high in the clouds calculates is the architecture that the service etc. of having merged network infrastructure and resource sharing is formed.In all services of sharing, internal memory and storage facilities are definitely the maximum projects of two demands.This is because the application of some hot topic, such as video streaming, googol is needed to store according to amount.When cloud service operates, internal memory and storage device management are very important, think that client maintains normal service quality.
For example, the server being used to provide cloud service usually manages or is connected on several rigid disk.Client uses this server, and data read from this rigid disk or write wherein.Zhao Yin in rigid disk system restriction and produce the delay of response time, the problem in demand for services can be caused.Under the operation of normal rigid disk system, when the access speed of application surface required (i.e. workload) exceed rigid disk system can provide time, thus the long response time produces.The bottleneck operated in the normally whole cloud service system of the ultimate load that rigid disk system can provide.In other words, the input-output operation number of times per second of rigid disk system cannot meet external need.For this problem, be necessary to remove or reduce workload to reach and to improve the usefulness of server.In implementation, the workload of part can be shared by other server (if any) or rigid disk, and those servers or rigid disk can add the existing rigid disk of support by automatic or manual wire over ground.No matter which kind of method above-mentioned is with solving this problem, its cost increased, and is in response to expecting a large amount of rigid disks of working condition and many deposits in advance, and the power consumption that necessity increases in order to extra computer hardware.From economic face, be really unworthy so doing.But the service level agreement due to system can specify the shortest time delay or minimum input-output operation number of times per second usually, therefore necessarily reaches.For maintain the operator of cloud service with limited fund for, how to reduce costs is an important problem.
It should be noted that the workload of server (rigid disk system) more or less can predict the differentiation in following a period of time according to historical record, the demand of cloud service workload is predictable.Therefore, can by reconfiguring at the intrasystem rigid disk of rigid disk, to reach the demand of workload by minimum cost.But a machine can not learn how and when to carry out reconfiguring of rigid disk.In many cases, this work is by authorized person, and come according to timely state or according to fixing program, it may not be fine for carrying out effect.
Another quick increased requirement the same with cloud service is that software definition stores.Software definition stores and refers to and can store the software of architecture from management, independently goes out to store the counter data storing technology of hardware.Under software definition stores, some function choosing-items can be started, as data de-duplication, copy, automatically simplify configuration, snapshot, and backup, tactical management is provided.By software definition storing technology, there is several front case can provide the solution of the problems referred to above.For example, in No. 20130297907th, U.S. Patent Publication, a kind of method for reconfiguring stocking system is disclosed.The method comprises two main steps: receive user to the demand information of storage device, and by the demand information of user, automatically produces the function setting of storage device and the device configuration file for this storage device; And use this function setting, be that one or more has the logical device of independent behavior feature automatically to reconfigure this storage device.The content of this application case indicates a kind of by software definition storage idea to reconfigure the new method of storage device.Method and system according to this application case also can allow the configuration of one or more logical device of user's dynamic conditioning, more flexibly to meet the demand information of user.But this application case but can not provide the change according to application surface demand (i.e. workload), can a kind of system of how storage device being reconfigured of automatic learning.
Therefore, the present invention discloses and a kind ofly stores the new system of automatic learning and resource re-allocation of realizing for software definition.System that employs adaptive control and operation, without the need to manual intervention.
Summary of the invention
Because prior art cannot provide stocking system according to the change of application surface demand, how automatic learning reconfigures its storage device, causes existing stocking system to be difficult to reach the requirement of service level agreement (ServiceLevelAgreement) or service quality (QualityofService) demand.Therefore inventor utilizes neural network algorithm, coordinate the operation of software definition stocking system, propose the invention solving foregoing problems.
According to a kind of aspect of the present invention, a kind of for software definition stocking system with the self-adaptation quick-response control system of improving SNR parameter, comprise: a flow monitoring module, is used to a storage node, obtain an observed value of performance parameter; The two neural module of one self-adaptation, in order between a particular value of this observed value and this performance parameter, in different difference situations, from storage device configuration historical record to relevant observed value, learn the best configuration of multiple storage device in this storage node, and when an existing difference is not less than a threshold value, provide this best configuration; And a rapid reaction control module, if this existing difference value is not less than this threshold value, in order to change in this storage node, the existing configuration of this storage device is the best configuration of storage device provided by the two neural module of this self-adaptation.The software that this storage node is stored by software definition operated, and after this best configuration adopts, this existing difference will reduce.
The two neural module of self-adaptation comprises: certain neural network assembly, and when this existing difference is not less than a permissible value, in order to provide those best configuration, and those best configuration are given tacit consent to before this self-adaptation quick-response control system runs; And an adaptive neural network assembly, in order in different difference situations, the historical record configured from this storage device and the relevant observed value in a long period, learn the best configuration of storage device in this storage node, and when existing difference is less than this permissible value but is not less than this threshold value, provide this best configuration.
According to this case conception, when this compose oneself operate through networking component time, this adaptive neural network assembly decommissions, and maybe when this adaptive neural network component operational, this composes oneself and to quit work through networking component.This permissible value is less than or equal to a default value, and this default value is 3 seconds.This long period scope is by during tens of second to whole historical record, and this observed value is not be recorded continuously in this long period.This variable quantity between best configuration and existing configuration provided through networking component that composes oneself, is greater than the variable quantity between best configuration and existing configuration provided by this adaptive neural network assembly.The best configuration learning this storage device reaches by neural network algorithm, and this particular value is the requirement of a service level agreement or a QoS requirement.This performance parameter is input-output operation number of times per second, time delay or circulation.This storage device is rigid disk, solid state hard disc, random access memory or its blend together combination, and this best configuration is the fixed qty that the number percent of different types storage device or unimodality storage device use.
This self-adaptation quick-response control system, comprises a computing module further, in order to the difference value that calculates this difference value and transmit this calculating to the two neural module of self-adaptation and rapid reaction control module.This traffic monitoring module, self-adaptation two neural module, rapid reaction control module or computing module are hardware, or the software that at least one processor in this storage node performs.
By above hardware implementing, system automatic learning can be allowed how to reconfigure its storage device, to reach the requirement of service level agreement or QoS requirement.
Accompanying drawing explanation
Fig. 1 illustrates the block scheme according to the self-adaptation quick-response control system of the embodiment of the present invention;
Fig. 2 shows the framework of a storage node;
Fig. 3 is the process flow diagram of the two neural module running of self-adaptation;
The best configuration table that Fig. 4 provides for the two neural module of self-adaptation.
Description of reference numerals: 10-self-adaptation quick-response control system; 100-storage node; 102-management server; 104-rigid disk; 106-solid state hard disc; 120-traffic monitoring module; 140-computing module; The two neural module of 160-self-adaptation; 162-composes oneself through networking component; 164-adaptive neural network assembly; 180-rapid reaction control module.Embodiment
The present invention more specifically describes with reference to following embodiment.
Refer to Fig. 1 to Fig. 4, be described according to one embodiment of the invention.Fig. 1 is according to the embodiment of the present invention, and one for the block scheme of the self-adaptation quick-response control system 10 of software definition stocking system.This system can be improved in a network, the performance parameter of software definition stocking system, such as input-output operation number of times per second, time delay or circulation.In the present embodiment, software definition stocking system is a storage node 100, and the time delay obtaining data from software definition stocking system illustrates as an example.This network can be the Internet.Thus, storage node 100 can be a database server, manages numerous storage facilitiess and provides client cloud service.It also can be a file server or mail server, has the storage facilities of exclusive use.This network also may be used for the LAN in laboratory, or for the wide area network of transnational enterprise, the present invention does not limit the application of storage node 100.But storage node 100 must be that software definition stores.In other words, the hardware (storage device) of storage node 100 should be able to be separated with the software of management storage node 100.The software that storage node 100 is stored by software definition operated.Therefore, in storage node 100, reshuffling of storage device can realize by other software each or hardware.
Ask for an interview Fig. 2, Fig. 2 shows the framework of storage node 100.Storage node 100 comprises 1 management server, 102,10 rigid disks 104, with 10 solid state hard discs 106.Management server 102 can receive instruction, to carry out reshuffling of rigid disk 104 and solid state hard disc 106.The difference configuration of storage node 100, the rigid disk 104 namely used and the number percent of solid state hard disc 106, can maintain certain time delay under different operating amount.Solid state hard disc 106 has than rigid disk 104 storage speed faster.But, under identical capacity, the price of solid state hard disc 106 expensive compared with rigid disk 104 go out many.Yi Yan's, same cost, the storage volume of rigid disk 104 is about ten times of solid state hard disc 106.For such storage node 100, because the life cycle of solid state hard disc 106 will decline quickly, providing complete will be uneconomic with the service that solid state hard disc 106 is standby, and when solid state hard disc 106 is all used, storage volume can become a problem.When the configuration packet of storage node 100 is containing some rigid disks 104 and solid state hard disc 106, as long as can meet the requirement of service level agreement (ServiceLevelAgreement) or service quality (QualityofService) demand time delay, this storage node 100 still can trouble-free operation and avoid aforesaid problem.
Self-adaptation quick-response control system 10 comprises the two neural module 160 of flow monitoring module 120, computing module 140, self-adaptation, with a rapid reaction control module 180.Traffic monitoring module 120 is used to the observed value obtaining storage node 100 time delay.Computing module 140 can calculate the difference between a particular value of an observed value and time delay, and the difference transmitting this calculating is to the two neural module 160 of self-adaptation and rapid reaction control module 180.Herein, the particular value of time delay refers to numerical value required in service level agreement or QoS requirement, it is storage node 100 the longest time delay, should carry out when super large workload occurs (make an exception when storage node 100 is started shooting or) in normal use in the lower service that provide.For the present embodiment, the particular value of time delay is 2 seconds.Any particular value is all feasible, and the present invention does not limit.
The two neural module 160 of self-adaptation is used to, in the historical record that configure from rigid disk 104 and solid state hard disc 106 and relevant observed value, when different difference, learn the best configuration of rigid disk 104 and solid state hard disc 106 in storage node 100.Difference is present between observed value and time delay particular value, and the two neural module 160 of self-adaptation also can provide best configuration to rapid reaction control module 180.When existing difference is not less than a threshold value, the two neural module 160 of self-adaptation just operates.So-called existing difference, refers to from the observed value of traffic monitoring module 120 and the up-to-date difference between particular value time delay (2 seconds).This threshold value exceeds the time for what time delay, particular value was preset.Because the time exceeding particular value time delay is too short, be just unworthy that the configuration changing rigid disk 104 and solid state hard disc 106 is to reduce time delay, existing configuration can continued operation.In the present embodiment, this threshold value is 0.2 second.Certainly, its different service that can provide for storage node 100 and changing.
In order to realize the running that the two neural module 160 of self-adaptation provides, the two neural module 160 of self-adaptation can comprise two main parts further: certain neural network assembly 162 and an adaptive neural network assembly 164.Composing oneself provides best configuration through networking component 162, and those best configuration are given tacit consent to before self-adaptation quick-response control system 160 runs, and when existing difference is not less than permissible value, composing oneself starts through networking component 162.Herein, this permissible value is the bonus values outside particular value time delay.Once this permissible value is aware, some emergency treatment must be carried out, and shortens time delay rapidly, so that client need not wait for the reply of storage node 100 within several seconds then always.The running composed oneself through networking component 162 can be regarded as the one restriction action of the time delay extended increasing with workload.In fact, this permissible value should be less than or equal a default value.Preferably be less than or equal 3 seconds.The present embodiment is using 3 seconds as this permissible value.
Adaptive neural network assembly 164 is used in different difference value situations, relevant observed value in the historical record configured from rigid disk 104 and solid state hard disc 106 and a long period, learns the best configuration of rigid disk 104 and solid state hard disc 106 in this storage node 100.It also can provide best configuration.When existing difference value is less than this permissible value but is not less than threshold value, adaptive neural network assembly 164 just operates.Aforesaid long period can be short to and arrive tens of second, during growing to the whole historical record of storage node 100.Can be provided the data acting on adaptive neural network assembly 164, to learn the best configuration of rigid disk 104 and solid state hard disc, any record of storage node 100 is all feasible.Preferably use the record data during whole historical record.Be understandable that in this long period, be not be recorded continuously some observation on value, some data also may be lost, but adaptive neural network assembly 164 still can use these discontinuous records.
Because the complexity of storage node 100 hardware and the different workload produced from the demand of client, by the time delay causing storage node 100 different, along with the time, there is not specific relation in time delay and workload.For self-adaptation quick-response control system 10, having the best mode of the management method of storage node 100 is lean on the variation relation self learnt therebetween.Therefore, neural network algorithm is a kind of good method reaching this target, and study rigid disk 104 can be reached by neural network algorithm with the best configuration of solid state hard disc 106.Although there are many neural network algorithms now, it is any that the present invention is not limited to use.In the pattern of each algorithm, the setting parameter of different layers can utilize the experience of other system to set up.
In order to know how the two neural module 160 of self-adaptation operates, and refers to Fig. 3, Fig. 3 is the process flow diagram that the two neural module 160 of self-adaptation operates.After the observed value of time delay is obtained by traffic monitoring module 120 (S01), and after the existing difference of this computing module 140 computing relay time (S02), the two neural module 160 of self-adaptation will judge whether that existing difference is not less than threshold value, 0.2 second (S03).If NO, rigid disk 104 and the existing configuration of solid state hard disc 106 remain unchanged (S04); If so, the two neural module 160 of self-adaptation will judge whether that existing difference is not less than this permissible value, 3 seconds (S05).If NO, adaptive neural network assembly 164 operates (S06); If so, compose oneself and to operate (S07) through networking component 162.It is obvious that when operating through networking component 162 when composing oneself, adaptive neural network assembly 164 decommissions; When adaptive neural network assembly 164 operates, composing oneself quits work through networking component 162.
If when existing difference is not less than threshold value, rapid reaction control module 180 to change in storage node 100 rigid disk 104 and the existing configuration of solid state hard disc 106, becomes the best configuration of rigid disk 104 that the two neural module 160 of self-adaptation provides and solid state hard disc 106.Therefore, the best configuration that rapid reaction control module 180 can always use the two neural module 160 of self-adaptation to provide, to adjust the configuration of storage node 100.Existing difference value, after best configuration adopts, can become less.
Ask for an interview the 4th figure, this is the best configuration table that the two neural module 160 of the present embodiment self-adaptation provides.When storage node 100 operates, when its time delay is less than 2 seconds, this configuration packet is containing the rigid disk 104 of 50% and the solid state hard disc 106 of 50%.Even if time delay, difference was in 0.2 second (time delay is 2.2 seconds), because time delay, difference was still less than threshold value, the two neural module 160 of self-adaptation can not operate, and configuration maintains former state.When time delay, difference value increased above 0.2 second, adaptive neural network assembly 164 operates, with the data of historical record and some new acceptance, the best configuration of study rigid disk 104 and solid state hard disc 106, the data of aforesaid new acceptance will be regarded as historical record for study.Meanwhile, based on the result of past study, when difference was not less than 0.2 second but was less than 0.5 second time delay, adaptive neural network assembly 164 provides the best configuration of rapid reaction control module 180 to be the rigid disk 104 of 40% and the solid state hard disc 106 of 60%; When difference was not less than 0.5 second but was less than 1.0 seconds time delay, best configuration is the rigid disk of 30% and the solid state hard disc 106 of 70%; When difference was not less than 1.0 seconds but was less than 3.0 seconds time delay, d best configuration is the rigid disk 104 of 20% and the solid state hard disc 106 of 80%.Certainly, because the behavior pattern of client can may change in future, best configuration from existing historical record study, can change further.After new best configuration was applied under difference different time delay, time delay is less than this particular value by becoming fast, 2 seconds.It should be noted that concerning best configuration, all point exponent numbers are not limited to said 6 above, can be to be greater than 6 or be less than 6.For example, time delay difference value sublevel quantity, can be 5 between threshold value and permissible value.In other words, within every 0.5 second, a sublevel is divided into.So, in the present embodiment, whole sublevel quantity becomes 8, but not 6.This is because learnt the best configuration come by the two neural module 160 of self-adaptation, rely on the kind of application (i.e. workload) demand, and the hardware specification of rigid disk and solid state hard disc in storage node 100.
When time delay, difference was not less than this permissible value, the appropriateness adjustment of configuration is late.In this case, a kind of compulsory means should be implemented to reduce time delay rapidly.Therefore, compose oneself and to operate and adaptive neural network assembly 164 decommissions through networking component 162.Composing oneself will provide the best configuration of acquiescence in rigid disk 104 and solid state hard disc 106 through networking component 162.According to the present embodiment, when difference was not less than 3.0 seconds but was less than 5.0 seconds time delay, best configuration is the rigid disk 104 of 10% and the solid state hard disc 106 of 90%; When time delay, difference was not less than 5.0 seconds, best configuration is the rigid disk 104 of 0% and the solid state hard disc 106 of 100%.In the case that this is extreme, employ all solid state hard discs 106.
But, although composing oneself can provide best configuration through networking component 162 and adaptive neural network assembly 164, can see from the 4th figure, the variable quantity between the best configuration provided through networking component 162 by composing oneself and existing configuration (rigid disk 104 of 50% and the solid state hard disc 106 of 50%) is greater than the variable quantity between best configuration and existing configuration provided by adaptive neural network assembly 164.
As mentioned above, time delay is only one of performance parameter of service level agreement requirement.Other performance parameter can change in the same way, changes that rigid disk 104 and solid state hard disc 106 configure to adjust.For example, input-output operation number of times per second and circulation can increase along with the increase of solid state hard disc 106.
It is emphasized that storage device is not limited to rigid disk and solid state hard disc, random access memory also can be used.Thus, rigid disk and random access memory, or the mixed form of taking of solid state hard disc and random access memory also can be employed.Best configuration in embodiment is the number percent of different types storage device in using.It also can be the fixed qty (as storage node only comprises solid state hard disc, reshuffle by increasing new or standby solid state hard disc and complete) that unimodality storage device uses.The most important thing is, the two neural module 160 of traffic monitoring module 120, computing module 140, self-adaptation, and rapid reaction control module 180 with hardware, or can perform by least one processor in storage node 100.
Although the present invention discloses as above with embodiment; so itself and be not used to limit the present invention; have in any art and usually know the knowledgeable; without departing from the spirit and scope of the present invention; when doing a little change and retouching, therefore protection scope of the present invention is when being as the criterion depending on the accompanying claim person of defining.
Claims (15)
1., for a self-adaptation quick-response control system for software definition stocking system, in order to improving SNR parameter, it is characterized in that, comprise:
One flow monitoring module, in order to by a storage node, obtains an observed value of performance parameter;
The two neural module of one self-adaptation, in order between a particular value of this observed value and this performance parameter, in different difference situations, from storage device configuration historical record to relevant observed value, learn the best configuration of multiple storage device in this storage node, and when an existing difference is not less than a threshold value, provide this best configuration; And
One rapid reaction control module, when this existing difference value is not less than this threshold value, in order to change in this storage node, the existing configuration of this storage device is the best configuration of storage device provided by the two neural module of this self-adaptation,
The software that wherein this storage node is stored by software definition operated, and after this best configuration adopts, this existing difference will reduce.
2. self-adaptation quick-response control system as claimed in claim 1, is characterized in that, the two neural module of this self-adaptation comprises:
Certain neural network assembly, when this existing difference is not less than a permissible value, in order to provide those best configuration, and those best configuration are given tacit consent to before this self-adaptation quick-response control system runs; And
One adaptive neural network assembly, in order in different difference situations, the historical record configured from this storage device and the relevant observed value in a long period, learn the best configuration of storage device in this storage node, and when existing difference value is less than this permissible value but is not less than this threshold value, provide this best configuration.
3. self-adaptation quick-response control system as claimed in claim 2, it is characterized in that, when this compose oneself operate through networking component time, this adaptive neural network assembly decommissions, maybe when this adaptive neural network component operational, this composes oneself and to quit work through networking component.
4. self-adaptation quick-response control system as claimed in claim 2, it is characterized in that, this permissible value is less than or equal to a default value.
5. self-adaptation quick-response control system as claimed in claim 4, it is characterized in that, this default value is 3 seconds.
6. self-adaptation quick-response control system as claimed in claim 2, it is characterized in that, this long period scope is by during tens of second to whole historical record.
7. self-adaptation quick-response control system as claimed in claim 2, it is characterized in that, this observed value is not be recorded continuously in this long period.
8. self-adaptation quick-response control system as claimed in claim 2, it is characterized in that, by this variable quantity between best configuration and existing configuration provided through networking component that composes oneself, be greater than the variable quantity between best configuration and existing configuration provided by this adaptive neural network assembly.
9. self-adaptation quick-response control system as claimed in claim 2, it is characterized in that, the best configuration learning this storage device reaches by neural network algorithm.
10. self-adaptation quick-response control system as claimed in claim 1, it is characterized in that, this particular value is the requirement of a service level agreement or a QoS requirement.
11. self-adaptation quick-response control systems as claimed in claim 1, is characterized in that, this performance parameter is input-output operation number of times per second, time delay or circulation.
12. self-adaptation quick-response control systems as claimed in claim 1, it is characterized in that, this storage device is rigid disk, solid state hard disc, random access memory or its blend together combination.
13. self-adaptation quick-response control systems as claimed in claim 1, is characterized in that, this best configuration is the number percent of different types storage device or the fixed qty of unimodality storage device use.
14. self-adaptation quick-response control systems as claimed in claim 1, is characterized in that, comprise a computing module further, in order to the difference that calculates this difference and transmit this calculating to the two neural module of self-adaptation and rapid reaction control module.
15. self-adaptation quick-response control systems as claimed in claim 1, it is characterized in that, this traffic monitoring module, self-adaptation two neural module, rapid reaction control module or computing module are hardware, or the software that at least one processor in this storage node performs.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410312060.7A CN105320461B (en) | 2014-07-01 | 2014-07-01 | Adaptive quick-response control system for software definition stocking system |
| HK16108996.9A HK1221035A1 (en) | 2014-07-01 | 2016-07-27 | Adaptive quick response controlling system for software defined storage system for improving performance parameter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410312060.7A CN105320461B (en) | 2014-07-01 | 2014-07-01 | Adaptive quick-response control system for software definition stocking system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105320461A true CN105320461A (en) | 2016-02-10 |
| CN105320461B CN105320461B (en) | 2018-04-03 |
Family
ID=55247921
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410312060.7A Active CN105320461B (en) | 2014-07-01 | 2014-07-01 | Adaptive quick-response control system for software definition stocking system |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN105320461B (en) |
| HK (1) | HK1221035A1 (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102035884A (en) * | 2010-12-03 | 2011-04-27 | 华中科技大学 | Cloud storage system and data deployment method thereof |
| US20120023219A1 (en) * | 2010-03-23 | 2012-01-26 | Hitachi, Ltd. | System management method in computer system and management system |
| CN102707995A (en) * | 2012-05-11 | 2012-10-03 | 马越鹏 | Service scheduling method and device based on cloud computing environments |
| CN103365781A (en) * | 2012-03-29 | 2013-10-23 | 国际商业机器公司 | Method and device for dynamically reconfiguring storage system |
-
2014
- 2014-07-01 CN CN201410312060.7A patent/CN105320461B/en active Active
-
2016
- 2016-07-27 HK HK16108996.9A patent/HK1221035A1/en not_active IP Right Cessation
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120023219A1 (en) * | 2010-03-23 | 2012-01-26 | Hitachi, Ltd. | System management method in computer system and management system |
| CN102035884A (en) * | 2010-12-03 | 2011-04-27 | 华中科技大学 | Cloud storage system and data deployment method thereof |
| CN103365781A (en) * | 2012-03-29 | 2013-10-23 | 国际商业机器公司 | Method and device for dynamically reconfiguring storage system |
| CN102707995A (en) * | 2012-05-11 | 2012-10-03 | 马越鹏 | Service scheduling method and device based on cloud computing environments |
Also Published As
| Publication number | Publication date |
|---|---|
| HK1221035A1 (en) | 2017-05-19 |
| CN105320461B (en) | 2018-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10733029B2 (en) | Movement of services across clusters | |
| EP3577561B1 (en) | Resource management for virtual machines in cloud computing systems | |
| US10055262B1 (en) | Distributed load balancing with imperfect workload information | |
| US10616370B2 (en) | Adjusting cloud-based execution environment by neural network | |
| US10243791B2 (en) | Automated adjustment of subscriber policies | |
| US9805140B2 (en) | Striping of directed graphs and nodes with improved functionality | |
| US10061233B2 (en) | Computer system backup performance optimization through performance analytics | |
| US20150317556A1 (en) | Adaptive quick response controlling system for software defined storage system for improving performance parameter | |
| US20220091763A1 (en) | Storage capacity forecasting for storage systems in an active tier of a storage environment | |
| US20180107503A1 (en) | Computer procurement predicting device, computer procurement predicting method, and recording medium | |
| EP3340534A1 (en) | A local sdn controller and corresponding method of performing network control and management functions | |
| US10171572B2 (en) | Server pool management | |
| US20150339587A1 (en) | Adaptive fuzzy rule controlling system for software defined storage system for controlling performance parameter | |
| US10409662B1 (en) | Automated anomaly detection | |
| CN105320461A (en) | Self-adaption fast reaction control system for software definition storage system | |
| US20150106820A1 (en) | Method and apparatus for providing allocating resources | |
| Montes et al. | Using global behavior modeling to improve qos in cloud data storage services | |
| US20180067659A1 (en) | Storage management device, performance adjustment method, and computer-readable recording medium | |
| TWI516956B (en) | Adaptive quick response controlling system for software defined storage system for improving performance parameter | |
| US10824413B2 (en) | Maintenance of computing nodes concurrently in a number updated dynamically | |
| CN105279560A (en) | Adaptive fuzzy rule control system used for software-defined storage system | |
| US10635334B1 (en) | Rule based data transfer model to cloud | |
| US11496499B2 (en) | Preventing scheduling or executing a resource on an inconsistent host node | |
| US11520638B1 (en) | Combined active and preinitialized resource management for rapid autoscaling | |
| US11481262B1 (en) | Rapid autoscaling with preinitialized instance quantity based on historical scale up rate |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1221035 Country of ref document: HK |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: GR Ref document number: 1221035 Country of ref document: HK |