CN106789726B - A kind of high concurrent service resource allocation method based on service priority - Google Patents
A kind of high concurrent service resource allocation method based on service priority Download PDFInfo
- Publication number
- CN106789726B CN106789726B CN201611187880.3A CN201611187880A CN106789726B CN 106789726 B CN106789726 B CN 106789726B CN 201611187880 A CN201611187880 A CN 201611187880A CN 106789726 B CN106789726 B CN 106789726B
- Authority
- CN
- China
- Prior art keywords
- service
- service request
- scheduler
- queue
- server
- 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
Classifications
-
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
- H04L47/2425—Traffic characterised by specific attributes, e.g. priority or QoS for supporting services specification, e.g. SLA
- H04L47/2433—Allocation of priorities to traffic types
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L47/00—Traffic control in data switching networks
- H04L47/10—Flow control; Congestion control
- H04L47/24—Traffic characterised by specific attributes, e.g. priority or QoS
- H04L47/2483—Traffic characterised by specific attributes, e.g. priority or QoS involving identification of individual flows
-
- 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
Abstract
The high concurrent service resource allocation method based on service priority that the present invention relates to a kind of, includes the following steps: to start;Scheduler receives the service request of user;Scheduler is that the service request received generates a process;Using the received service request of polling mode forwarding;When there is server to receive the service request, terminate;When Servers-all all suspends access request, the service request that the unserviced device of user receives is stored into insertion respective queue;It is retransmitted every time t using polling mode according to the time interval t that priority level initializing retransmits the service request queue, until there is server free, which is received, and is terminated.While this method guarantees resource energy relatively uniform distribution, realize relatively simple;When Service Source is in short supply, the high business of priority can preferentially get Service Source.
Description
Technical field
The present invention relates to service resource allocation technical field more particularly to a kind of high concurrent services based on service priority
Resource allocation methods.
Background technique
In recent years, with the continuous development of computer technology and the communication technology, Internet number of users exponentially increases again,
Internet application has obtained rapid proliferation.The progress of technology excites the growth of user demand, and business also becomes more complicated, this
Server is required to provide the ability of high concurrent access and reasonable distribution Service Source.
Traditional individual server processing capacity is limited, can not handle high concurrent service request, and by single common clothes
It is again excessively high that business device is replaced as its cost of the server of higher performance, and causes the waste of idle server resource again.Therefore
Multiple servers are configured to a server cluster system, are jointly processed by the way of external request, it has also become domestic and international structure
Build the main method of high-performance server.
Group system is compared with using traditional individual server more complex, how for service request reasonable distribution cluster
In Service Source the problem of being most critical, it will directly affect whether group system can be provided equal or close to expected property
Energy.Such as when largely requesting while reaching, if the unreasonable certain server node that will lead to of service resource allocation is chronically at
Overload state, and in addition then the long period is in idle condition certain server.The above situation serious waste collection
The resources advantage of group's system, the request of user also cannot be responded timely, and also there is no much for the overall performance of group system
Promotion, this with introduce group system original intention be not consistent.Therefore a kind of preferable load balancing scheme must be used, with
Just each server node task being reasonably assigned to as far as possible in group system.
Load-balancing technique occurs along with the appearance of Clustering, it is therefore an objective to equably be scheduled to request in cluster
Node, accomplish load balancing as far as possible.Currently used load-balancing algorithm is divided into following two categories:
(1) static scheduling algorithm
Static scheduling algorithm time of occurrence is earliest, is basis and the blank of other algorithms.Its basic thought is exactly fixed in advance
The better scheduling rule of justice, all requests are allocated according to rule, and the current state with system, such as load capacity,
The information such as resource utilization are unrelated.
(2) dynamic dispatching algorithm
For static scheduling, dynamic dispatching algorithm joined some feedback mechanisms, and this mechanism can be dynamically
Current cluster information is collected, load including each node, type, quantity, the utilization rate of bandwidth, request delay newly requested
Etc..Using these information, we can dynamically change the weight and other some attributes of server, and what change newly connected walks
To.Such measure load summit and it is unstable when, good adjustment effect can be played, avoid service tilt and delay
The case where machine, occurs, and improves the overall performance of cluster.
Since dynamic dispatching algorithm can dynamically change resource allocation policy according to the real time information of system, in reality scene
In have better performance, the current country is related in the patent of load-balancing technique, using the more of dynamic dispatching algorithm.
But in the prior art about the angle that the scheme of distribution Service Source is mostly from load balancing, very
The rare Resource Allocation Formula for service priority.In the case where Service Source is abundant, since all business can be timely
It is assigned to Service Source, priority may be less for the importance of business.However, when Service Source is in short supply, it is not all
Business can be assigned to Service Source in time, Service Source just should be preferentially assigned to for core, urgent business, this
When, priority for business importance with regard to self-evident.
Summary of the invention
In view of above-mentioned analysis, the present invention is intended to provide a kind of high concurrent service resource allocation side based on service priority
Method, this method fully consider service priority, according to priority distribute Service Source, right in conventional load balancing technique to solve
Service priority considers insufficient problem.
The purpose of the present invention is mainly achieved through the following technical solutions:
Scheduler is arranged in a kind of high concurrent service resource allocation method based on service priority;The scheduler and service
Servers-all in device cluster is connected;The scheduler is stored with the mark of Servers-all, forms the clothes of the server cluster
Business device identifies queue;
The high concurrent service resource allocation method includes the following steps:
Step S1, scheduler receive the service request that user issues;
Step S2, the scheduler are that each service request received generates a process;
Step S3, the scheduler using the received service request of polling mode forwarding to server, when there is service
Device Service Source is sufficient, and when receiving the service request, corresponding process terminates;When Servers-all all Service Source deficiencies, temporarily
When stopping access request, it is transferred to step S4;
Step S4, the scheduler store the service request that unserviced device receives;
Step S5, the service priority of scheduler service request of calling stored from configuration file;
Step S6, the time is issued according to service priority and service request, the service request of storage is inserted by scheduler
Corresponding position in corresponding service request queue;
Specifically, service request queue is made of the identical service request of priority, in each service request queue, is pressed
The chronological order that user issues service request is lined up;
Step S7, the scheduler is according to the service priority in service request queue and the industry in the service request queue
The time interval t that the service request queue is retransmitted in business quantity setting specifically meets niti<njtj, wherein i, j indicate that business is asked
Queue label is sought, i label service request queue service priority is higher than the service priority of j label service request queue, ni、ti
Service request quantity respectively in the queue of i label service request and the time interval for retransmitting the service request queue, nj、tjPoint
Service request quantity in j label service request queue and the time interval of the service request queue Wei not be retransmitted;
Step S8, the scheduler are that every service request queue generates a process;
Step S9, the scheduler retransmits corresponding service request queue every time t, using polling mode, until this
All service requests of service request queue are received by server, terminate the process.
Further, the step S3 includes following sub-step:
Step S301, scheduler Service Source units needed for obtaining the service request in configuration file;Specifically, it uses
Service Source units needed for each type of service is stored in configuration file by family, is scheduled for device acquisition;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S302, the scheduler;
Received service request and the required Service Source units got are transmitted to by step S303, the scheduler
The server of tail of the queue mark;
Step S304, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to tune
Spend the required Service Source units of device forwarding;It is that the server receives and handle the service request, terminates the process;It is no, it should
Server returns to " business pause access " response to scheduler;
Repeat the above steps S301~S304, until there is a server to receive the service request, terminates corresponding process or institute
There is server to return to " business pause access " response, goes to step S4.
Further, the service priority is determined by user demand, urgent, significance level of the user according to task, setting
Service priority is written in configuration file, is scheduled for device and transfers by service priority.
Further, the step S9 includes following sub-step:
Step S901, scheduler are provided from service needed for first service request of the service request queue is obtained in configuration file
Source units;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S902, scheduler;
After step S903, t time interval, the scheduler is by first service request and the industry in the service request queue
Business requests required Service Source units to be transmitted to the server that tail of the queue identifies;
Step S904, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to tune
Spend the required resource units number of device forwarding;It is that the server receives and handle first service request, and scheduler should
Service request is deleted from business request queue;No, which returns to " business pause access " response to scheduler;
Repeat the above steps S901~S904, until all service requests in the service request queue are connect by server
By terminating the process.
Further, scheduler uses two-server to realize with active/standby mode.
The present invention is in order to solve the above technical problems, provide another high concurrent Service Source based on service priority point
Method of completing the square, specifically:
Scheduler is set;The scheduler is connected with the Servers-all in server cluster;The scheduler is stored with all
The mark of server forms the server identification queue of the server cluster;Scheduler is at regular intervals to Servers-all
It inquires Service Source Expenditure Levels, obtains the total resources consumption ratio of server cluster;
The high concurrent service resource allocation method includes the following steps:
Step S1, scheduler receive the service request that user issues;
Step S2, scheduler judge that whether the total resources consumption of current server cluster than reaches threshold set by scheduler
Value;Judging result is when being, to go to step S5;When judging result is no, S3 is gone to step;
Step S3, the scheduler are that each service request received generates a process;
Step S4, the scheduler using the received service request of polling mode forwarding to server, when there is service
Device Service Source is sufficient, and when receiving the service request, corresponding process terminates;
Step S5, the scheduler store the service request;
Step S6, the service priority of scheduler service request of calling stored from configuration file;Business is excellent
First grade is determined that user is arranged service priority, service priority is write according to urgent, the significance level of task by user demand
Enter in configuration file, is scheduled for device and transfers;
Step S7, the time issued according to service priority and service request, which is inserted into corresponding service
Corresponding position in request queue;
Specifically, service request queue is made of the identical service request of priority, in each service request queue, is pressed
The chronological order that user issues service request is lined up;
Step S8, the scheduler is according to the service priority in service request queue and the industry in the service request queue
The time interval t that the service request queue is retransmitted in business quantity setting specifically meets niti<njtj, wherein i, j indicate that business is asked
Queue label is sought, i service request queue service priority is higher than the service priority of j service request queue, ni、tiRespectively i industry
Service request quantity in business request queue and the time interval of the service request queue is retransmitted, nj、tjRespectively j service request
Service request quantity in queue and the time interval for retransmitting the service request queue;
Step S9, the scheduler are that every service request queue generates a process;
Step S10, the scheduler retransmits corresponding service request queue every time t, using polling mode, until this
All service requests of service request queue are received by server, stop retransmitting the service request queue, terminate the process;
Further, the threshold value is determined by user demand, is arranged according to the actual situation, and setting principle is with excellent to business
First grade stresses the increase of degree, reduces the threshold value.
Further, the threshold value is taken as 75%.
Further, the step S4 includes following sub-step:
Step S401, scheduler Service Source units needed for obtaining the service request in configuration file;Specifically, it uses
Service Source units needed for each type of service is stored in configuration file by family, is scheduled for device acquisition;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S402, the scheduler;
Step S403, the required Service Source units that the scheduler will acquire are transmitted to the service of tail of the queue mark
Device;
Step S404, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to tune
Spend the required Service Source units of device forwarding;It is that the server receives and handle the service request, terminates the process, it is no, it should
Server returns to " business pause access " response to scheduler;
Repeat the above steps S401~S404, until there is a server to receive the service request.
Further, the step S10 includes following sub-step:
Step S1001, scheduler service needed for obtaining first service request of the service request queue in configuration file
Resource units number;
After step S1002, t time interval, team's head element in server identification queue is gone out team and is inserted into team by scheduler
Tail;
Step S1003, the scheduler is by Service Source unit needed for first service request in the service request queue
Number is transmitted to the server of tail of the queue mark;
Step S1004, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to
The required resource units number of scheduler forwarding;It is that the server receives and first business in processing business request queue is asked
It asks, corresponding service request is deleted from business request queue;No, then the server returns to " business pause access " to scheduler
Response;
Repeat the above steps S1001~S1004, until all service requests in the service request queue are by server
Receive, stops retransmitting the service request queue, terminate the process.
The present invention has the beneficial effect that:
1, present invention setting scheduler is high concurrent traffic assignments Service Source, to each business in high concurrent business
Request generates a process, one service request of each process independent process.To any bar service request by the way of poll
Service Source is distributed, while guaranteeing resource energy relatively uniform distribution, is realized relatively simple.
2, the present invention devises a kind of mechanism that service resource allocation is realized according to service priority, ensure that and provides in service
When source is in short supply, the high business of priority can preferentially get Service Source.
3, the present invention realizes service priority by the size of the time interval of service request repeating transmission, the reality of the priority
Existing mode ensure that the business of different priorities can get Service Source within a certain period of time, be not in that priority is low
Business can not all obtain the case where Service Source within a very long time, therefore be very suitable for the scene of high concurrent business
4, by setting threshold value, the allocation strategy based on priority is opened when thinking necessary by user, is guaranteed
In Service Source relative deficiency, the business for keeping priority high preferentially gets Service Source.
Other features and advantages of the present invention will illustrate in the following description, also, partial become from specification
It obtains it is clear that understand through the implementation of the invention.The objectives and other advantages of the invention can be by written explanation
Specifically noted structure is achieved and obtained in book, claims and attached drawing.
Detailed description of the invention
Attached drawing is only used for showing the purpose of specific embodiment, and is not to be construed as limiting the invention, in entire attached drawing
In, identical reference symbol indicates identical component.
Fig. 1 is one flow diagram of the embodiment of the present invention.
Fig. 2 is three flow diagram of the embodiment of the present invention.
Specific embodiment
Specifically describing the preferred embodiment of the present invention with reference to the accompanying drawing, wherein attached drawing constitutes the application a part, and
Together with embodiments of the present invention for illustrating the principle of the present invention.
Embodiment one,
Scheduler is arranged before carrying out service resource allocation in the present embodiment;Scheduler can be in server cluster
Any server can be individually scheduled, can also processing business request simultaneously.Institute in the scheduler and server cluster
There is server to be connected;The scheduler is stored with the mark of Servers-all, forms the server identification queue of the server cluster.
Scheduler is realized using a high performance network server with muti-piece network interface card.In order to improve fault-tolerance, using active/standby mode,
Realize that dual-locomotive heat back-up (is simultaneously deployed to set of system on two-server, if host system goes out with two-server
Existing failure, then be immediately switched to run on standby host).
The high concurrent service resource allocation method based on service priority that present embodiment discloses a kind of, as shown in Figure 1,
Include the following steps:
Step S1, scheduler receive the service request that user issues;
Step S2, the scheduler are that each service request received generates a process;
Step S3, the scheduler using the received service request of polling mode forwarding to server, when there is service
Device Service Source is sufficient, and when receiving the service request, corresponding process terminates;When Servers-all all Service Source deficiencies, temporarily
When stopping access request, it is transferred to step S4;
Including following sub-step:
Step S301, scheduler Service Source units needed for obtaining the service request in configuration file;Specifically, it uses
Service Source units needed for each type of service is stored in configuration file by family, is scheduled for device acquisition;
Demand of the different business to Service Source is different, assigns a Service Source units, such as a industry to every kind of business
Be engaged in it is smaller to resources requirement, assign the Service Source of its 1 unit, b business is slightly larger to Service Source demand, assign its 2
The resource of a unit concurrently sets the total resource units number of server, and when being set as 10000 units, server can be simultaneously
The most 10000 a service requests of concurrent processing perhaps 5000 b service requests or 8000 a business and 1000 b business.
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S302, the scheduler;
Received service request and the required Service Source units got are transmitted to by step S303, the scheduler
The server of tail of the queue mark;
Step S304, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to tune
Spend the required Service Source units of device forwarding;It is that the server receives and handle the service request, terminates the process, it is no, it should
Server returns to " business pause access " response to scheduler;
Repeat the above steps S301~S304, and until there is a server to receive the service request, corresponding process terminates;Or institute
There is server to return to " business pause access " response, is transferred to step S4.
Step S4, the scheduler store the service request that unserviced device receives;
Step S5, the service priority of scheduler service request of calling stored from configuration file;Business is excellent
First grade is determined by user demand, and according to urgent, the significance level of task, service priority is arranged, and service priority write-in is matched
It sets in file, is scheduled for device and transfers;
When Service Source is relatively abundant, different business is not obvious the contention of resource because these business can and
When be assigned to Service Source, but when Service Source is in short supply, existing resource can not meet all business in time, this
When, the priority of resource allocation is just particularly important.Business urgent for those, important, core needs to assign
Their relatively high priority guarantee that they are preferentially assigned to resource, and not urgent, secondary, non-core for those
Business can assign their relatively low priority.
Step S6, the time is issued according to service priority and service request, the service request of storage is inserted by scheduler
Corresponding position in corresponding service request queue;
Specifically, service request queue is made of the identical service request of priority, in each service request queue, is pressed
The chronological order that user issues service request is lined up;
Step S7, scheduler is according to the service priority in service request queue and the business number in the service request queue
The time interval t that the service request queue is retransmitted in amount setting specifically meets niti<njtj, wherein i, j indicate service request team
Column label, i label service request queue service priority are higher than the service priority of j label service request queue, ni、tiRespectively
For the service request quantity in i label service request queue and retransmit the time interval of the service request queue, nj、tjRespectively j
Service request quantity in label service request queue and the time interval for retransmitting the service request queue;
Step S8, the scheduler are that every service request queue generates a process;
Step S9, the scheduler retransmits corresponding service request queue every time t, using polling mode, until this
All service requests of service request queue are received by server, terminate the process;Including following sub-step:
Step S901, scheduler are provided from service needed for first service request of the service request queue is obtained in configuration file
Source units;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S902, scheduler;
After step S903, t time interval, the scheduler is by first service request and the industry in the service request queue
Business requests required Service Source units to be transmitted to the server that tail of the queue identifies;
Step S904, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to tune
Spend the required resource units number of device forwarding;It is that the server receives and handle first service request, and scheduler should
Service request is deleted from business request queue;No, which returns to " business pause access " response to scheduler;
Repeat the above steps S901~S904, until all service requests in the service request queue are connect by server
By terminating the process.
The working condition of server, which is divided into, in the present embodiment cluster loads less than, two states of full load.It loads less than, refers to
Be in cluster there are also not up to full load server.If individual server has reached the peak load of work, i.e., full
Load, will refuse new request, when server all runnings at full capacity all in cluster, then entire cluster is in full load
State can not handle new request.When cluster load is less than, priority is not distinguished between different business, i.e., scheduler is with identical
Priority is that different business distributes Service Source.At this point, scheduler uses the mode of polling dispatching for each traffic assignments clothes
Business resource.If response that all servers all return in dispatcher queue " business pause access ", illustrates server set at this time
Group has been in full load state, then enters the full load stage.When cluster is in full load state, Service Source has been in deficient
Weary state, the request of scheduler are illustrated not available in cluster by after Servers-all response " business pause access "
Service Source.
The present embodiment thinks highly of the frequency (i.e. time interval) of sending service request by configuration scheduling to control Service Source
Distribution.Specifically, when scheduler discovery server cluster has reached full load, current service request can temporarily be stored
Get off, t retransmits the request at regular intervals, just stops retransmitting until the request is received.At this point, if there is several inhomogeneities
The business of type is in contention Service Source, in order to allow more urgent, important, core business to be preferentially quickly distributed to service money
Source can retransmit scheduler the high of the frequency setting of the type service request, i.e., the time interval t's for retransmitting the request
Value is set smaller.
Embodiment two,
For ease of understanding, one is provided specifically with example.
In core communication system, the calculating service provided using calculation server cluster is required there are three types of business,
Priority is respectively crucial control business, operational administrative business, general inquiry business from high to low.
Assuming that calculation server cluster is made of 10 single calculation servers, and externally provide access controller
ServController is responsible for receiving the business of forwarding user terminal by ServController as unified access point, that is, scheduler
It requests to be handled to calculation server cluster.When calculation server cluster operational capability is still available free, for three kinds of business
Computing resource request, ServController calls 10 calculation servers with equal priority, using polling mode.It is elected
The calculation server calculated load selected has expired, then the calculation server returns to " business pause access " to ServController and rings
It answering, ServController is used according to calculation server next in return value selective listing, and so on, when in list
All calculation servers all polls one time and in the case where all returning to " business pause access ", ServController is temporarily deposited
The business request information is stored up, reselects calculation server after being spaced n seconds time.
For all calculation servers all in the case where at full capacity, ServController activates preferential fraction automatically at this time
With strategy, preferentially got by setting the size of different business retransmittal interval n automatically to control the business of high priority
Computing resource is handled.Crucial control business n=1s, operational administrative business n=2s, general inquiry business n=5s are such as set,
When calculation server collection group operatione at full capacity when, three kinds of service requests are all responded " business pause access ", at this point,
ServController retransmits three kinds of service requests again after 1s, 2s, 5s respectively, after 1s, any one calculation server from
Become idle at full capacity, then crucial control business obtains at first calculates Service Source;After 2s, another calculation server is from completely
Load becomes idle, and exists without the business of higher priority of resource allocation, and operational administrative business can be obtained meter at this time
Calculate Service Source.Obviously, in this way, the higher business of priority of resource allocation can preferentially be obtained in contention for resources
Service Source, the low business of priority will not can not all obtain Service Source within a very long time.
Embodiment three,
Above-described embodiment one, two in server cluster full load, is just opened based on priority distribution service money
Source present embodiments provides a kind of preferential based on threshold value unlatching to make urgent, important business obtain service resources faster
The example of grade distribution Service Source.
The high concurrent service resource allocation method based on service priority that present embodiment discloses a kind of, scheduler is every one
It fixes time and inquires Service Source Expenditure Levels to Servers-all, the total resources consumption ratio of server cluster is obtained, when total money
When threshold of the source consumption than reaching scheduler setting, illustrate that Service Source is not abundant enough at this time, in order to allow urgent business
Preferentially get resource, the method for salary distribution regardless of priority mean allocation resource before stopping, but first forwarding priority compared with
High service request, then the lower service request of forwarding priority.
As shown in Fig. 2, including the following steps:
Step S1, scheduler receive the service request that user issues;
Step S2, scheduler judge that whether the total resources consumption of current server cluster than reaches threshold set by scheduler
Value;Judging result is when being, to go to step S5;When judging result is no, S3 is gone to step;The threshold value determines by user demand, according to
Actual conditions setting, with the increase for stressing degree to service priority, reduces the threshold value;Optimal threshold is 75%, that is, is existed
Resource consumption ratio reach 75% or more Service Source it is relatively nervous when open priority allocation strategy, dispatched with guaranteeing to increase
Device program considers that priority distributes the equilibrium of both Service Sources.
Step S3, the scheduler are that each service request received generates a process;
Step S4, the scheduler using the received service request of polling mode forwarding to server, when there is service
Device Service Source is sufficient, and when receiving the service request, corresponding process terminates;
Including following sub-step:
Step S401, scheduler Service Source units needed for obtaining the service request in configuration file;Specifically, it uses
Service Source units needed for each type of service is stored in configuration file by family, is scheduled for device acquisition;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S402, the scheduler;
Step S403, the required Service Source units that the scheduler will acquire are transmitted to the service of tail of the queue mark
Device;
Step S404, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to tune
Spend the required Service Source units of device forwarding;It is that the server receives and handle the service request, terminates the process, it is no, it should
Server returns to " business pause access " response to scheduler;
Repeat the above steps S401~S404, until there is a server to receive the service request.
Step S5, the scheduler store the service request;
Step S6, the service priority of scheduler service request of calling stored from configuration file;Business is excellent
First grade is determined that user is arranged service priority, service priority is write according to urgent, the significance level of task by user demand
Enter in configuration file, is scheduled for device and transfers;
Step S7, the time issued according to service priority and service request, which is inserted into corresponding service
Corresponding position in request queue;
Specifically, service request queue is made of the identical service request of priority, in each service request queue, is pressed
The chronological order that user issues service request is lined up;
Step S8, scheduler is according to the service priority in service request queue and the business number in the service request queue
The time interval t that the service request queue is retransmitted in amount setting specifically meets niti<njtj, wherein i, j indicate service request team
Column label, i label service request queue service priority are higher than the service priority of j label service request queue, ni、tiRespectively
For the service request quantity in i label service request queue and retransmit the time interval of the service request queue, nj、tjRespectively j
Service request quantity in label service request queue and the time interval for retransmitting the service request queue;
Step S9, the scheduler are that every service request queue generates a process;
Step S10, the scheduler retransmits corresponding service request queue every time t, using polling mode, until this
All service requests of service request queue are received by server, stop retransmitting the service request queue, terminate the process;Including
Following sub-step:
Step S1001, scheduler service needed for obtaining first service request of the service request queue in configuration file
Resource units number;
After step S1002, t time interval, team's head element in server identification queue is gone out team and is inserted into team by scheduler
Tail;
Step S1003, the scheduler is by Service Source unit needed for first service request in the service request queue
Number is transmitted to the server of tail of the queue mark;
Step S1004, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to
The required resource units number of scheduler forwarding;It is that the server receives and first business in processing business request queue is asked
It asks, corresponding service request is deleted from business request queue;No, then the server returns to " business pause access " to scheduler
Response;
Repeat the above steps S1001~S1004, until all service requests in the service request queue are by server
Receive, stops retransmitting the service request queue, terminate the process.
The invention has the following advantages that
1, the present invention uses the mode of poll for high concurrent traffic assignments Service Source, is guaranteeing that resource can be relatively uniform point
With while, realize it is relatively simple.
2, the present invention devises a kind of resource allocation mechanism by service priority, ensure that relatively in short supply in Service Source
When, the high business of priority can preferentially get Service Source.
3, the present invention realizes service priority by the size of the time interval of service request repeating transmission, the reality of the priority
Existing mode ensure that the business of different priorities can get Service Source within a certain period of time, be not in that priority is low
Business can not all obtain the case where Service Source within a very long time, therefore be very suitable for the scene of high concurrent business.
It will be understood by those skilled in the art that realizing all or part of the process of above-described embodiment method, meter can be passed through
Calculation machine program is completed to instruct relevant hardware, and the program can be stored in computer readable storage medium.Wherein, institute
Stating computer readable storage medium is disk, CD, read-only memory or random access memory etc..
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
In the technical scope disclosed by the present invention, any changes or substitutions that can be easily thought of by anyone skilled in the art,
It should be covered by the protection scope of the present invention.
Claims (10)
1. a kind of high concurrent service resource allocation method based on service priority, it is characterised in that: setting scheduler;The scheduling
Device is connected with the Servers-all in server cluster;The scheduler is stored with the mark of Servers-all, forms the server
The server identification queue of cluster;
The high concurrent service resource allocation method includes the following steps:
Step S1, scheduler receive the service request that user issues;
Step S2, the scheduler are that each service request received generates a process;
Step S3, the scheduler using the received service request of polling mode forwarding to server, when there is server clothes
Resource of being engaged in is sufficient, and when receiving the service request, corresponding process terminates;When Servers-all all Service Sources are insufficient, pause is connect
When entering to request, it is transferred to step S4;
Step S4, the scheduler store the service request that unserviced device receives;
Step S5, the service priority of scheduler service request of calling stored from configuration file;Service priority
It is determined by user demand, service priority is arranged according to urgent, the significance level of task in user, and service priority write-in is matched
It sets in file, is scheduled for device and transfers;
Step S6, the time is issued according to service priority and service request, the service request of storage is inserted into accordingly by scheduler
Service request queue in corresponding position;
Specifically, service request queue is made of the identical service request of priority, in each service request queue, by user
The chronological order for issuing service request is lined up;
Step S7, the scheduler are asked according to the service priority in service request queue and the business in the service request queue
The time interval t for asking quantity setting to retransmit the service request queue specifically meets niti<njtj, wherein i, j indicate that business is asked
Queue label is sought, i service request queue service priority is higher than the service priority of j service request queue, ni、tiRespectively i industry
Service request quantity in business request queue and the time interval of the service request queue is retransmitted, nj、tjRespectively j service request
Service request quantity in queue and the time interval for retransmitting the service request queue;
Step S8, the scheduler are that every service request queue generates a process;
Step S9, the scheduler retransmits corresponding service request queue every time t, using polling mode, until the business
All service requests of request queue are received by server, terminate the process.
2. high concurrent service resource allocation method according to claim 1, it is characterised in that: the step S3 includes as follows
Sub-step:
Step S301, scheduler Service Source units needed for obtaining the service request in configuration file;Specifically, user will
Service Source units needed for each type of service is stored in configuration file, is scheduled for device acquisition;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S302, the scheduler;
Received service request and the required Service Source units got are transmitted to tail of the queue by step S303, the scheduler
The server of mark;
Step S304, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to scheduler
The required Service Source units of forwarding;It is that the server receives and handle the service request, terminates the process, no, the service
Device returns to " business pause access " response to scheduler;
Repeat the above steps S301~S304, and until there is a server to receive the service request, corresponding process terminates;Or it is all
Server returns to " business pause access " response, is transferred to step S4.
3. high concurrent service resource allocation method according to claim 1, it is characterised in that: the service priority by with
Family demand determines, according to urgent, the significance level of task, service priority is arranged, service priority is written in configuration file,
Device is scheduled for transfer.
4. high concurrent service resource allocation method according to claim 1, it is characterised in that: the step S9 includes as follows
Sub-step:
Step S901, scheduler Service Source list needed for obtaining first service request of the service request queue in configuration file
Digit;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S902, scheduler;
After step S903, t time interval, the scheduler asks first service request and the business in the service request queue
Required Service Source units is asked to be transmitted to the server of tail of the queue mark;
Step S904, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to scheduler
The required resource units number of forwarding;It is that the server receives and handle first service request, scheduler is by the business
Request is deleted from business request queue;No, which returns to " business pause access " response to scheduler;
Repeat the above steps S901~S904, until all service requests in the service request queue are received by server,
Terminate the process.
5. high concurrent service resource allocation method according to claim 1, it is characterised in that: the scheduler uses two
Server is realized with active/standby mode.
6. a kind of high concurrent service resource allocation method based on service priority, it is characterised in that: setting scheduler;The scheduling
Device is connected with the Servers-all in server cluster;The scheduler is stored with the mark of Servers-all, forms the server
The server identification queue of cluster;Scheduler inquires Service Source Expenditure Levels to Servers-all at regular intervals, obtains
The total resources of server cluster consume ratio;
The high concurrent service resource allocation method includes the following steps:
Step S1, scheduler receive the service request that user issues;
Step S2, scheduler judge that whether the total resources consumption of current server cluster than reaches threshold value set by scheduler;
Judging result is when being, to go to step S5;When judging result is no, S3 is gone to step;
Step S3, the scheduler are that each service request received generates a process;
Step S4, the scheduler using the received service request of polling mode forwarding to server, when there is server clothes
Resource of being engaged in is sufficient, and when receiving the service request, corresponding process terminates;
Step S5, the scheduler store the service request;
Step S6, the service priority of scheduler service request of calling stored from configuration file;Service priority
It is determined by user demand, service priority is arranged according to urgent, the significance level of task in user, and service priority write-in is matched
It sets in file, is scheduled for device and transfers;
Step S7, the service request is inserted into corresponding service request by the time issued according to service priority and service request
Corresponding position in queue;
Specifically, service request queue is made of the identical service request of priority, in each service request queue, by user
The chronological order for issuing service request is lined up;
Step S8, the scheduler are asked according to the service priority in service request queue and the business in the service request queue
The time interval t for asking quantity setting to retransmit the service request queue specifically meets niti<njtj, wherein i, j indicate that business is asked
Queue label is sought, i service request queue service priority is higher than the service priority of j service request queue, ni、tiRespectively i industry
Service request quantity in business request queue and the time interval of the service request queue is retransmitted, nj、tjRespectively j service request
Service request quantity in queue and the time interval for retransmitting the service request queue;
Step S9, the scheduler are that every service request queue generates a process;
Step S10, the scheduler retransmits corresponding service request queue every time t, using polling mode, until the business
All service requests of request queue are received by server, stop retransmitting the service request queue, terminate the process;
7. high concurrent service resource allocation method according to claim 6, it is characterised in that: the threshold value is by user demand
It determines, is arranged according to the actual situation, setting principle is to reduce the threshold value with the increase for stressing service priority degree.
8. high concurrent service resource allocation method according to claim 7, it is characterised in that: the threshold value is taken as 75%.
9. high concurrent service resource allocation method according to claim 6, it is characterised in that: the step S4 includes as follows
Sub-step:
Step S401, scheduler Service Source units needed for obtaining the service request in configuration file;Specifically, user will
Service Source units needed for each type of service is stored in configuration file, is scheduled for device acquisition;
Team's head element in server identification queue is gone out team and is inserted into tail of the queue by step S402, the scheduler;
Step S403, the required Service Source units that the scheduler will acquire are transmitted to the server of tail of the queue mark;
Step S404, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to scheduler
The required Service Source units of forwarding;It is that the server receives and handle the service request, terminates the process;It is no, the service
Device returns to " business pause access " response to scheduler;
Repeat the above steps S401~S404, until there is a server to receive the service request, terminates corresponding process.
10. high concurrent service resource allocation method according to claim 6, it is characterised in that: the step S10 includes such as
Lower sub-step:
Step S1001, scheduler Service Source needed for obtaining first service request of the service request queue in configuration file
Units;
After step S1002, t time interval, team's head element in server identification queue is gone out team and is inserted into tail of the queue by scheduler;
Step S1003, the scheduler turn Service Source units needed for first service request in the service request queue
Issue the server of tail of the queue mark;
Step S1004, the server of the tail of the queue mark judge whether self residual Service Source units is more than or equal to scheduling
The required resource units number of device forwarding;It is that the server receives and first service request in processing business request queue, phase
The service request answered is deleted from business request queue;No, then the server returns to " business pause access " response to scheduler;
Repeat the above steps S1001~S1004, until all service requests in the service request queue are connect by server
By the service request queue is retransmitted in stopping, terminating the process.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611187880.3A CN106789726B (en) | 2016-12-20 | 2016-12-20 | A kind of high concurrent service resource allocation method based on service priority |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201611187880.3A CN106789726B (en) | 2016-12-20 | 2016-12-20 | A kind of high concurrent service resource allocation method based on service priority |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106789726A CN106789726A (en) | 2017-05-31 |
CN106789726B true CN106789726B (en) | 2019-09-10 |
Family
ID=58894324
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201611187880.3A Active CN106789726B (en) | 2016-12-20 | 2016-12-20 | A kind of high concurrent service resource allocation method based on service priority |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106789726B (en) |
Families Citing this family (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107230026A (en) * | 2017-06-21 | 2017-10-03 | 安徽瑞弋自动化科技有限公司 | A kind of industrial equipment polling control method |
CN107633451A (en) * | 2017-10-23 | 2018-01-26 | 深圳市中润四方信息技术有限公司 | A kind of tax-related service processing method, system |
CN107948095B (en) * | 2017-11-21 | 2021-11-02 | 中国银行股份有限公司 | Resource control method and device and bus system server |
CN108124003B (en) * | 2017-12-11 | 2021-03-26 | 中盈优创资讯科技有限公司 | Network management equipment connection processing method, device and system |
CN108111499B (en) * | 2017-12-15 | 2021-07-16 | 腾讯科技(深圳)有限公司 | Business processing performance optimization method and device, electronic equipment and storage medium |
CN108510093A (en) * | 2018-03-29 | 2018-09-07 | 重庆柚瓣家科技有限公司 | Community service reservation system |
CN108573313A (en) * | 2018-03-29 | 2018-09-25 | 重庆柚瓣家科技有限公司 | Intensive service deployment method for family endowment |
CN108303688B (en) * | 2018-04-27 | 2022-02-11 | 北京东远润兴科技有限公司 | Reconstruction system and method for radar signal processing and radar system |
CN110971533B (en) * | 2018-09-30 | 2023-02-03 | 京东方科技集团股份有限公司 | Method, server apparatus, client apparatus, and medium for data communication |
US10911568B2 (en) * | 2018-10-03 | 2021-02-02 | Twitter, Inc. | Client software back off |
CN109492926A (en) * | 2018-11-22 | 2019-03-19 | 中国银行股份有限公司 | A kind of transaction dispatcher method and system based on preferential level adaptation |
CN110086848B (en) * | 2019-03-15 | 2023-07-21 | 安徽银柏电子科技有限公司 | Network request sending method and device, computer equipment and storage medium |
CN111026555B (en) * | 2019-12-23 | 2023-11-24 | 北京数衍科技有限公司 | Priority level flow separation compensation method and device and electronic equipment |
CN111737000A (en) * | 2020-06-22 | 2020-10-02 | 四川长虹电器股份有限公司 | Method for realizing load balance |
CN113239061B (en) * | 2021-05-31 | 2023-02-10 | 浙江环玛信息科技有限公司 | Intelligent court data updating method and system |
CN113467909B (en) * | 2021-06-29 | 2022-03-15 | 贝壳找房(北京)科技有限公司 | Time consuming method and apparatus for compressing concurrent requests |
CN116208680B (en) * | 2023-05-04 | 2023-07-14 | 成都三合力通科技有限公司 | Server access management system and method |
CN117112239B (en) * | 2023-10-23 | 2024-02-09 | 合肥综合性国家科学中心人工智能研究院(安徽省人工智能实验室) | Extensible load balancing method and system on heterogeneous reasoning back end |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102497421A (en) * | 2011-12-12 | 2012-06-13 | 湖南科技大学 | Web server cluster system round-robin scheduling load balancing method with good fairness |
CN103428870A (en) * | 2012-05-22 | 2013-12-04 | 中兴通讯股份有限公司 | Method and device for determining priority of access service |
CN104202261A (en) * | 2014-08-27 | 2014-12-10 | 华为技术有限公司 | Service request processing method and device |
CN104243405A (en) * | 2013-06-08 | 2014-12-24 | 华为技术有限公司 | Request processing method, device and system |
-
2016
- 2016-12-20 CN CN201611187880.3A patent/CN106789726B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102497421A (en) * | 2011-12-12 | 2012-06-13 | 湖南科技大学 | Web server cluster system round-robin scheduling load balancing method with good fairness |
CN103428870A (en) * | 2012-05-22 | 2013-12-04 | 中兴通讯股份有限公司 | Method and device for determining priority of access service |
CN104243405A (en) * | 2013-06-08 | 2014-12-24 | 华为技术有限公司 | Request processing method, device and system |
CN104202261A (en) * | 2014-08-27 | 2014-12-10 | 华为技术有限公司 | Service request processing method and device |
Also Published As
Publication number | Publication date |
---|---|
CN106789726A (en) | 2017-05-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106789726B (en) | A kind of high concurrent service resource allocation method based on service priority | |
CN107066319B (en) | Multi-dimensional scheduling system for heterogeneous resources | |
CN103856393B (en) | Distributed message middleware system and its operation method based on database | |
CN1744575B (en) | Method for improving quality-of-service management in networks | |
JP5324637B2 (en) | Dynamic flowlet scheduling system, flow scheduling method, and flow scheduling program | |
EP2944072B1 (en) | System for service allocation | |
CN101873005B (en) | Method for realizing balanced acquisition of electric energy | |
CN108616458A (en) | The system and method for schedule packet transmissions on client device | |
US20130290526A1 (en) | Network assisted power management | |
CN101257445A (en) | Method for synchronous traffic shaping | |
CN103079273A (en) | Resource management method suitable for multilayer satellite system | |
CN103927225A (en) | Multi-core framework Internet information processing and optimizing method | |
CN109088829A (en) | A kind of data dispatching method, device, storage medium and equipment | |
Khalifa¹ et al. | Collaborative autonomic resource management system for mobile cloud computing | |
CN112565774A (en) | Video transcoding resource scheduling method and device | |
CN109783225A (en) | A kind of tenant's priority management method and system of multi-tenant big data platform | |
CN109962760A (en) | A kind of business scheduling method suitable for wireless TDMA ad hoc network | |
CN105515791A (en) | Ethernet method and use method in ethernet device | |
CN108282526B (en) | Dynamic allocation method and system for servers between double clusters | |
Wen et al. | Load balancing job assignment for cluster-based cloud computing | |
CN107122235A (en) | Public infrastructure resource regulating method based on application priority | |
Lu et al. | An efficient load balancing algorithm for heterogeneous grid systems considering desirability of grid sites | |
CN101867580B (en) | Method for allocating network flow and device | |
CN105873223A (en) | Service scheduling method, service scheduling device, wireless network controller and base station | |
CN109714223B (en) | System and method for realizing network service access dynamic load sharing function under NFV architecture |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |