CN108494588A - A kind of system and method for cluster block device dynamic QoS configuration - Google Patents
A kind of system and method for cluster block device dynamic QoS configuration Download PDFInfo
- Publication number
- CN108494588A CN108494588A CN201810202165.5A CN201810202165A CN108494588A CN 108494588 A CN108494588 A CN 108494588A CN 201810202165 A CN201810202165 A CN 201810202165A CN 108494588 A CN108494588 A CN 108494588A
- Authority
- CN
- China
- Prior art keywords
- qos
- performance
- lun
- node
- cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5003—Managing SLA; Interaction between SLA and QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/50—Network service management, e.g. ensuring proper service fulfilment according to agreements
- H04L41/5003—Managing SLA; Interaction between SLA and QoS
- H04L41/5019—Ensuring fulfilment of SLA
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/1008—Server selection for load balancing based on parameters of servers, e.g. available memory or workload
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/10—Protocols in which an application is distributed across nodes in the network
- H04L67/1001—Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
- H04L67/1004—Server selection for load balancing
- H04L67/101—Server selection for load balancing based on network conditions
Abstract
The present invention discloses a kind of system and method for cluster block device dynamic QoS configuration, which includes：Several sub-module storeds；And performance monitoring submodule, QoS submodules, the clustering performance balanced device of configuration corresponding with each sub-module stored；This real-time Transmission is given the clustering performance balanced device elected, clustering performance balanced device to be responsible for balancing cluster QoS settings, and distributes the QoS settings of each node by the performance information of each each node of performance monitoring submodule real-time collecting；Each QoS submodules receive the QoS setting informations that the clustering performance balanced device elected issues, and apply it to each node.Technical solution using the present invention, no matter how client with cluster establishes service connection, and no matter how business IO changes, and can accomplish the reasonable distribution of QoS settings.
Description
Technical field
The present invention relates to cluster-based storage administrative skill field, more particularly to a kind of cluster block device dynamic QoS configuration is
System and method.
Background technology
Traditional block device QoS technology depends on the Cgroup mechanism of Linux bottoms, the disadvantage is that QoS is arranged just for list
Node, single logical unit number (LUN, Logical Unit Number, refer in particular to specified block device) are come into force.For cluster environment,
One block device can go out from multiple heads, and traditional QoS mechanism drawback appears, and can not be competent at the application environment of cluster complexity.
100M is read for example, for one needing that single LUN is arranged, writes the QoS configurations of 100M, traditional scheme is applied in cluster
Two kinds of situations in environment：
(1) referring to Fig. 1, for LUN1, it is each store head be respectively arranged 100M tape reading is wide and 100M writes bandwidth, in this way
For the angle of cluster, LUN1 actually has the tape reading of 300M wide and 300M writes bandwidth.It is this to be mismatched with demand, setting
It is unreasonable.
(2) referring to Fig. 2, consider that each storage head QoS that be averaged is arranged, each head reading 33M writes 33M.This set pair
Compare bandwidth demand that mean time is preferable in each client, consider demand of the actual conditions client to bandwidth may and it is uneven
, for example some period Client2 does not have business, and at this moment the total bandwidth of cluster only reads 66M for the customer, writes
66M.There are still mismatched with demand, unreasonable situation is arranged.
From being analyzed above it can be seen that be can not be according to the service conditions of client, each the drawbacks of tradition QoS schemes
QoS settings are dynamically adjusted between a node, result in the unreasonable distribution of supply and demand.Therefore, the prior art is there are problem, need into
One step is improved.
Invention content
In view of the problems of the existing technology, the present invention provides system and the side of a kind of cluster block device dynamic QoS configuration
Method.
First, the present invention provides a kind of system of cluster block device dynamic QoS configuration, and specific technical solution is as follows：
A kind of system of cluster block device dynamic QoS configuration, including：Several sub-module storeds；And with each storage submodule
Block corresponds to the performance monitoring submodule, QoS submodules, clustering performance balanced device of configuration；Each performance monitoring submodule real-time collecting
The performance information of each node gives this real-time Transmission the clustering performance balanced device elected, clustering performance balanced device to be responsible for flat
The cluster QoS that weighs is arranged, and distributes the QoS settings of each node；Each QoS submodules receive under the clustering performance balanced device elected
The QoS setting informations of hair, apply it to each node.
Preferably, the clustering performance balanced device summarizes the performance of each LUN in cluster wide using tree data structure
The performance information of information and each of the links.
Preferably, the performance information includes that LUN tape reading is wide, iops, LUN write bandwidth, LUN writes iops for LUN readings, average IO
Delay.
The present invention also provides a kind of method of cluster block device dynamic QoS configuration, specific technical solution is as follows.
A kind of method of cluster block device dynamic QoS configuration, includes the following steps：
The performance information of S1, each each node of performance monitoring submodule real-time collecting；
S2, the clustering performance balanced device real-time reception performance information elected are formed each after balancing cluster QoS settings
The QoS setting informations of node；
S3, each QoS submodules receive corresponding QoS setting informations, apply it to each node.
Preferably, performance information described in step S1 includes that LUN tape reading is wide, iops, LUN write bandwidth, LUN writes for LUN readings
Iops, average IO delays.
Preferably, clustering performance balanced device described in step S2 is summarized each in cluster wide using tree data structure
The performance information of LUN and the performance information of each of the links.
Preferably, the method for the setting of clustering performance balanced device balancing cluster QoS described in step S2 is：Count assigned I/O class
Then the head number of type is calculated and is averaged.
Technical solution using the present invention, no matter how client with cluster establishes service connection, no matter business IO how
Variation can accomplish the reasonable distribution of QoS settings.
Description of the drawings
Fig. 1 is prior art diagram of the present invention；
Fig. 2 is another prior art diagram of the present invention；
Fig. 3 is the schematic diagram of the system of dynamic QoS of the present invention configuration；
Fig. 4 is the flow chart of the method for dynamic QoS of the present invention configuration.
Specific implementation mode
Below in conjunction with the drawings and specific embodiments, the present invention is further described.
With reference to Fig. 3, the present invention provides a kind of system of cluster block device dynamic QoS configuration, including：Several storage submodules
Block；And performance monitoring submodule, QoS submodules, the clustering performance balanced device of configuration corresponding with each sub-module stored；Each performance
This real-time Transmission is given the clustering performance balanced device elected, collection by the performance information of each node of monitoring submodule real-time collecting
Group's balancing performance device is responsible for balancing cluster QoS settings, and distributes the QoS settings of each node；Each QoS submodules reception elects
The QoS setting informations that issue of clustering performance balanced device, apply it to each node.
Specifically, a performance is increased for the sub-module stored of each node based on original cluster block device system
Monitoring submodule, a QoS submodules, a clustering performance balanced device need not appoint client (Client) in the framework
What is changed, Client application unaware.Wherein：
Performance monitoring submodule --- the performance information of real-time collecting present node, including but not limited to LUN tape reading width,
Iops, LUN write bandwidth, LUN writes iops for LUN readings, average IO delays, per second that information is real-time transmitted to the sociability elected
It can balanced device.
QoS submodules --- it is responsible for receiving the QoS setting informations that clustering performance balanced device issues, and is applied to and works as prosthomere
Point.
Clustering performance balanced device --- it is responsible for collecting the performance data of each memory node of cluster, inside uses tree shaped data
Structure, which summarizes each performance data of LUN and the performance data of each of the links, each memory node in cluster wide, to be run
One clustering performance balanced device, clustering performance balanced device plays most critical in entire framework, and each node in cluster can
A performance balanced device program is run, synchronization indeed is in operating status there are one clustering performance balanced device to be entire
Cluster service.It is to ensure the high reliability of service to design multiple, and after the delay machine of a node, other clustering performances are equal
Weighing apparatus can seize service, but only can seize success there are one clustering performance balanced device.
With reference to figure 3 and Fig. 4, the present invention also provides a kind of methods of cluster block device dynamic QoS configuration, including walk as follows
Suddenly：
The performance information of S1, each each node of performance monitoring submodule real-time collecting；
S2, the clustering performance balanced device real-time reception performance information elected are formed each after balancing cluster QoS settings
The QoS setting informations of node；
S3, each QoS submodules receive corresponding QoS setting informations, apply it to each node.
In step s 2, the clustering performance balanced device elected is responsible for balancing cluster QoS settings and distributes each head
QoS setting.QoS settings are managed as unit of LUN, and definition tape reading width is rdkbps, and it is rdiops to read io number, writes band
Width is wrkbps, and it is wriops to write io number, and each index can be respectively set.In preferred embodiment, the sociability
Can the method for balanced device balancing cluster QoS setting be：Statistics specifies the head number of I/O type, then calculates and averages.Below
Specifically introduce the equalization algorithm.
Assuming that in moment t, LUN1 has X head to write IO, and Y head has reading IO.Then current Real-time ORB computational methods
For：
Table 1
| The interim QoS settings of head | There is reading business head | Without reading business head |
| rdkbps | Qos.rdkbps/X | Without limitation |
| rdiops | Qos.rdiops/X | Without limitation |
| wrkbps | Qos.wrkbps/Y | Without limitation |
| wriops | Qos.wriops/Y | Without limitation |
QoS balanced devices use a kind of simple dependable algorithm, count the head number for specifying I/O type, then calculate
It averages.It is all greedy because of major applications to select this algorithm, and storage load of going to save that can't be specially occupies.
Storage end can not be expected type of service simultaneously, so it is also a kind of best approach ensureing justice to average.
In the present invention, QoS settings can be adjusted with business change dynamic：
(1), there are four types of service conditions for moment t heads, are reading and writing, read-write respectively, are write without reading-nothing；
(2), moment t to t+1 is in case of business change, and alteration has 12 kinds, such as following table：
Table 2
| The T moment | The T+1 moment |
| It reads | Read-write |
| It reads | It writes |
| It reads | No reading-nothing is write |
| It writes | Read-write |
| It writes | It reads |
| It writes | No reading-nothing is write |
| Read-write | It reads |
| Read-write | It writes |
| Read-write | No reading-nothing is write |
| No reading-nothing is write | It reads |
| No reading-nothing is write | It writes |
| No reading-nothing is write | Read-write |
Balanced device can count within every 3 seconds primary each LUN each head (node that storage access is provided in cluster) industry
Business situation.If there is head service conditions changes, then interim QoS can be recalculated at once, and notify to need machine to be changed
Head is arranged using new QoS.
Design final effect is no matter how client with cluster establishes service connection in this way, no matter how business IO becomes
Change, can always accomplish the reasonable distribution of QoS settings.
The foregoing is merely the preferred embodiment of the present invention, are not intended to limit the scope of the invention, every at this
Under the inventive concept of invention, using equivalent structure transformation made by description of the invention and accompanying drawing content, or directly/use indirectly
In the scope of patent protection that other related technical areas are included in the present invention.
Claims (7)
1. a kind of system of cluster block device dynamic QoS configuration, which is characterized in that including：Several sub-module storeds；And with it is each
Sub-module stored corresponds to the performance monitoring submodule, QoS submodules, clustering performance balanced device of configuration；Each performance monitoring submodule
The performance information of each node of real-time collecting gives this real-time Transmission the clustering performance balanced device elected, clustering performance balanced
Device is responsible for balancing cluster QoS settings, and distributes the QoS settings of each node；Each QoS submodules receive the clustering performance elected
The QoS setting informations that balanced device issues, apply it to each node.
2. the system of dynamic QoS configuration according to claim 1, which is characterized in that the clustering performance balanced device uses
Tree data structure summarizes each performance information of LUN and the performance information of each of the links in cluster wide.
3. the system of dynamic QoS configuration according to claim 1, which is characterized in that the performance information includes LUN tape reading
Width, iops, LUN write bandwidth, LUN writes iops for LUN readings, average IO delays.
4. a kind of method of cluster block device dynamic QoS configuration, which is characterized in that include the following steps：
The performance information of S1, each each node of performance monitoring submodule real-time collecting；
S2, the clustering performance balanced device real-time reception performance information elected form each node after balancing cluster QoS settings
QoS setting informations；
S3, each QoS submodules receive corresponding QoS setting informations, apply it to each node.
5. the method for dynamic QoS configuration according to claim 4, which is characterized in that performance information packet described in step S1
Include that LUN tape reading is wide, LUN reads that iops, LUN write bandwidth, LUN writes iops, average IO delays.
6. the method for dynamic QoS configuration according to claim 4, which is characterized in that clustering performance is equal described in step S2
Weighing apparatus summarizes each performance information of LUN and the performance information of each of the links in cluster wide using tree data structure.
7. the method for dynamic QoS configuration according to claim 4, which is characterized in that clustering performance is equal described in step S2
The method of weighing apparatus balancing cluster QoS setting is：Statistics specifies the head number of I/O type, then calculates and averages.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810202165.5A CN108494588A (en) | 2018-03-12 | 2018-03-12 | A kind of system and method for cluster block device dynamic QoS configuration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810202165.5A CN108494588A (en) | 2018-03-12 | 2018-03-12 | A kind of system and method for cluster block device dynamic QoS configuration |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108494588A true CN108494588A (en) | 2018-09-04 |
Family
ID=63339001
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810202165.5A Pending CN108494588A (en) | 2018-03-12 | 2018-03-12 | A kind of system and method for cluster block device dynamic QoS configuration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108494588A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111131390A (en) * | 2019-11-25 | 2020-05-08 | 深圳市瑞云科技有限公司 | Storage caching method for improving cloud rendering concurrency number |
| CN111245740A (en) * | 2019-11-19 | 2020-06-05 | 华为技术有限公司 | Service quality strategy method and device for configuration service and computing equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102426552A (en) * | 2011-10-31 | 2012-04-25 | 成都市华为赛门铁克科技有限公司 | Storage system service quality control method, device and system |
| CN103164169A (en) * | 2011-12-19 | 2013-06-19 | 富士通株式会社 | Storage device, and storage device control method |
| CN103763130A (en) * | 2013-12-31 | 2014-04-30 | 华为数字技术（苏州）有限公司 | Method, device and system for managing large-scale cluster |
-
2018
- 2018-03-12 CN CN201810202165.5A patent/CN108494588A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102426552A (en) * | 2011-10-31 | 2012-04-25 | 成都市华为赛门铁克科技有限公司 | Storage system service quality control method, device and system |
| CN103164169A (en) * | 2011-12-19 | 2013-06-19 | 富士通株式会社 | Storage device, and storage device control method |
| CN103763130A (en) * | 2013-12-31 | 2014-04-30 | 华为数字技术（苏州）有限公司 | Method, device and system for managing large-scale cluster |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111245740A (en) * | 2019-11-19 | 2020-06-05 | 华为技术有限公司 | Service quality strategy method and device for configuration service and computing equipment |
| CN111245740B (en) * | 2019-11-19 | 2022-06-28 | 华为云计算技术有限公司 | Service quality strategy method and device for configuration service and computing equipment |
| CN111131390A (en) * | 2019-11-25 | 2020-05-08 | 深圳市瑞云科技有限公司 | Storage caching method for improving cloud rendering concurrency number |
| CN111131390B (en) * | 2019-11-25 | 2022-06-21 | 深圳市瑞云科技有限公司 | Storage caching method for improving cloud rendering concurrency number |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7039685B2 (en) | Traffic measurement methods, devices, and systems | |
| US7643983B2 (en) | Data storage system emulation | |
| US7035971B1 (en) | Request scheduling to mirrored heterogeneous storage arrays | |
| US9535866B2 (en) | Asymmetric storage device wide link | |
| US6820172B2 (en) | Method, system, and program for processing input/output (I/O) requests to a storage space having a plurality of storage devices | |
| US20060235901A1 (en) | Systems and methods for dynamic burst length transfers | |
| JP6179321B2 (en) | Storage management device, control method, and control program | |
| US20060294045A1 (en) | Control of service workload management | |
| CN108494588A (en) | A kind of system and method for cluster block device dynamic QoS configuration | |
| US20120324160A1 (en) | Method for data access, message receiving parser and system | |
| US20070168565A1 (en) | Storage control system and method | |
| CN105430103A (en) | Dynamic load balancing system based on multi-controller storage | |
| CN109933431A (en) | A kind of intelligent client load equalization methods and system | |
| US10657045B2 (en) | Apparatus, system, and method for maintaining a context stack | |
| CN107506145A (en) | A kind of physical store dispatching method and cloud main frame creation method | |
| CN104580018A (en) | Bandwidth adjustment method and device in software-defined network | |
| US9075728B2 (en) | Disk array device and method for controlling disk array device | |
| CN108123963B (en) | API auxiliary system and processing method for cross-border e-commerce | |
| CN100505680C (en) | Network traffic control system | |
| US10200269B2 (en) | Information generation method using write speed of data to storage device, information generation device using write speed of data to storage device, and non-transitory recording medium for storing information generation program using write speed of data to storage device | |
| CN105024875A (en) | Method and system for measuring network flow rate of broadband classified users | |
| JP2013005448A (en) | System and method for calculating utilization entropy | |
| CN106990919A (en) | The memory management method and device of automatic separating fault disk | |
| US7117118B2 (en) | Efficient accumulation of performance statistics in a multi-port network | |
| US20110023046A1 (en) | Mitigating resource usage during virtual storage replication |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180904 |