CN1237767C - A resource access shared scheduling and controlling method and apparatus - Google Patents
A resource access shared scheduling and controlling method and apparatus Download PDFInfo
- Publication number
- CN1237767C CN1237767C CN 200410062763 CN200410062763A CN1237767C CN 1237767 C CN1237767 C CN 1237767C CN 200410062763 CN200410062763 CN 200410062763 CN 200410062763 A CN200410062763 A CN 200410062763A CN 1237767 C CN1237767 C CN 1237767C
- Authority
- CN
- China
- Prior art keywords
- request
- priority
- time
- resource
- requests
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 230000008569 process Effects 0.000 claims abstract description 49
- 238000012545 processing Methods 0.000 claims abstract description 36
- 230000015572 biosynthetic process Effects 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000012163 sequencing technique Methods 0.000 claims description 4
- 230000004044 response Effects 0.000 abstract description 8
- 230000001276 controlling effect Effects 0.000 description 12
- 238000005755 formation reaction Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 230000000737 periodic effect Effects 0.000 description 6
- 230000005540 biological transmission Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 230000032683 aging Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 238000011112 process operation Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000012546 transfer Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003139 buffering effect Effects 0.000 description 1
- 239000007795 chemical reaction product Substances 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Landscapes
- Computer And Data Communications (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a shared resource access scheduling and controlling method and an apparatus, which belong to the technical field of network computers. In the method, resource requests from MMNC are received according to arriving sequences; after the resource requests are classified, the resource requests are added to tail parts of queues with corresponding priorities a multipriority request queue; a scheduling apparatus divides the processing time of a processor into fixed scheduling periods, and each scheduling period is divided into processing time sections according to the priorities; besides, a certain time slice quota is assigned to each time section. In one scheduling period, firstly, the requests are selected according to the service conditions of the time slice quotas; secondly, if all the time slice quotas are used up, but the scheduling processing period still has residual time slices, the requests are selected according to the priorities of the requests; finally, if free time slices still exist after the previous three sorts of requests are all processed, pre-operating requests are selected for processing. The present invention can effectively enhance the response speed of real-time requests and as well as process resource requests of low priorities.
Description
Technical field
The invention belongs to the network computer technical field, particularly a kind of managing under the multi-media network computer distributed system environment, the concurrent access request of shared resource on its server is carried out the method and the device of scheduling controlling.
Background technology
Along with the develop rapidly of hardware and software technology, it is more extensive that the use of PC is just becoming.According to the statistics of International Data Corporation (IDC) (IDC), China's PC sales volume in 2002 has reached 1,120 ten thousand, and the application quantity of PC is also being pressed annual speed increase more than 20%.On the other hand, PC also exposes many problems when function is from strength to strength, higher relatively as price, the maintenance and management complexity, and the user uses inconvenience, poor stability etc.In the network calculations epoch, people attempt to adopt the diverse network end product that utilizes network technology to replace or part replaces PC.Since the nineties in 20th century, some big computer vendors are set about low-cost, the high performance network computer of development one after another both at home and abroad, and oneself the solution of having released one after another, as NC, NetPC, no dish station, WBT etc.
The present application people has declared " a kind of network computer of local no operating system " (patent No. is ZL03122022.3).Different with the terminal computation schema that with WBT (Windows Based Terminal) is representative, this network computer adopts server to store and management, client are calculated and the computation schema of man-machine interaction.This network computer itself does not have mass-memory unit, and installing operating system and operate in types of applications program on the operating system not on it has only one section guidance code to leave among the ROM.After powering up, this guidance code with start server and carry out a series of alternately, according to operational environment the demand of system is loaded corresponding operating system; When the new application function of needs, client computer is called in the application program on the server local computing environment load and execution as required, and the shared resource on the access server, or even the information resources on the Internet is simultaneously finished various calculating or information and obtained task.The network computer of employing aforementioned calculation pattern in use user need not carry out any configuration, plug and play; The upgrading of its employed operating system and application software, configuration, maintenance and management all carry out on server, thereby have really realized client plug and play and zero management.In addition, adopt the network computer of aforementioned calculation pattern to overcome present thin client to the undue shortcoming that relies on of server, client CPU not only is responsible for receiving user's input in this locality and graphical interfaces is handled, also carry out actual computing, round Realization and compatible PC desktop application function comprise complicated multimedia function.This network computer comprises: network computer mainboard and corresponding external equipment thereof, said mainboard comprises low power consumption CPU, internal memory and other companion chip, said mainboard also comprises a remote activation chip that is used for the long-range loading of operating system, after the long-range loading of operating system, utilize one or more technology accesses among the loading as required of NFS, program, Distributed Calculation or the terminal technology to leave file, program or data on the server in again, realize the desktop function of conventional P C.
Program loads as required and is meant that network computer leaves multipad on the server in, in running, and the program that only loads the user as required and asked, other program still leaves on the server.When moving the application of forming by a plurality of program files, also be only subprogram file load wherein to be arrived local execution as required.
In the present invention, adopt the network computer of aforementioned calculation pattern to be referred to as to manage multi-media network computer (MMNC, Manageable Multimedia Network Computer).Can manage the multi-media network computer and adopt Ethernet to be connected with server, server provides resource (comprising text data, multi-medium data, application program, result of calculation etc.) share service for many network computers.
The server that MMNC connects mainly is general PC, and when a plurality of MMNC visited the PC server, server was extremely important to the efficient of access control and response.NFS (NFS, NetworkFile System) and CIFS (CIFS, Common Internet File System) etc. are adopted in existing shared resource visit.In these methods, server is put on an equal footing all concurrent requests, adopts first service (FIFO) tactful sequential processes earlier.This scheduling mode lays particular emphasis on research and how to reduce expense and effectively utilize the network bandwidth, thereby improves the average response time of entire system, and does not consider the ageing and service quality QoS of single request.
Summary of the invention
The present invention has proposed a kind of dispatch control method and device of the MMNC of being used for server shared resource visit for addressing the above problem, the efficient when being used to improve and improve visit MMNC server.Server can adopt corresponding strategy to handle at the different timeliness demands of request, shorten the request responding time that needs are responded fast, and provide the application service quality support for managing the multi-media network computer.
The invention provides a kind of dispatch control method that is used for shared resource visit on the server---based on the periodic scheduling algorithm (PPSA of priority, Priority-based Periodic Scheduling Algorithm), this method is based on 2 points: promptly the request of using is carried out scheduling based on priority by the qos requirement classification with to these classification requests.PPSA appends to the progress information that MMNC uses in the resource request, and server is classified to the resource request of network application according to different network applications, the stage of client process operation and the type that is requested resource, gives different priority; And be the cycle with the empirical value, in the mode of time-slotting the request of different priorities is periodically dispatched, for MMNC is provided by the better response service that provides.Specifically, server is put into the reception formation according to the resource request of the sequencing reception that arrives from a plurality of MMNC.Not only comprise required resource information in the resource request, also comprise the application message of the request of sending simultaneously and the demand (service quality QoS) of request response timeliness.Immediately monitor by application state MMNC, and in conjunction with the additional information in the resource request, server will be mapped to different stage to the timeliness demand of resource request, and resource request classified, for it demarcates different priority, and add the subqueue of corresponding priority in the multipriority queue.Server is in the process of handling resource request, and PPSA adopts the periodicity strategy based on priority that resource request is carried out scheduling controlling.PPSA is divided into the processor time dispatching cycle of regular length, PPSA gives all kinds of requests with the timeslice quota according to certain pro rate in dispatching cycle, form the privately owned processing time section of all kinds of requests, the timeslice quota in the privately owned time period preferentially is used for corresponding class processing of request.In a scheduling processing cycle, if in the time of will from the multipriority request queue, choosing a resource request, if with there being a plurality of resource request that are positioned at different priorities request queue head to satisfy schedulable condition, and the processing time section of corresponding each priority query all also remains time enough sheet quota, then the resource request in the priority treatment high-priority queue; If the resource request of certain priority request formation head satisfies schedulable condition, but the processing time sheet quota of this priority correspondence uses up, even then its priority is high again, also must hand to the lower priority request that corresponding processing time section still has sheet quota remaining time to disposal right; In addition, if certain class request processing time section has timeslice quota more than needed, but No Assets request temporarily in the formation of corresponding priority, then remaining timeslice quota can be by the resource request utilization of all the other priority, and wherein the higher priority class resource request preferentially obtains handling.
According to said method, the present invention also provides a kind of scheduling controlling device that is used for the visit of shared resource on the server, and this device comprises: one receives the formation parts, is used for sequencing according to the time of advent and receives resource request from MMNC; One client application information table parts are used to write down MMNC and go up the relevant information of using; One grader according to the relevant information of using, is classified to the resource request that receives in the team successively, and it is divided into plurality of classes, and is the different priority of giving of all categories; One multipriority request queue parts comprise a plurality of subqueues unit, and wherein the corresponding priority in each subqueue unit joins in the corresponding priority subqueue unit through the sorted resource request of grader; One scheduler is used for choosing resource request from described multipriority request queue parts, and handles.
Characteristics of the present invention and effect:
The present invention compares with pure priority scheduling mode with using the FIFO processing mode, improves a lot for real-time request responding speed, and can take into account the low priority resources processing of request simultaneously, avoids it to be blocked by high priority requests for a long time.
The scheduling controlling device and method of shared resource disclosed in this invention visit not only is applicable to the MMNC server, and be widely used in other adopt customer end/server mode, server need carry out the situation of scheduling controlling to the shared resource visit of the information terminal received.
Description of drawings
Fig. 1 is the server operational environment schematic diagram of the embodiment of the invention;
Fig. 2 is the workflow schematic diagram of the access control apparatus of the embodiment of the invention;
Fig. 3 be in the access control apparatus of the embodiment of the invention MMNC application information table data structure diagram;
Fig. 4 is the schematic diagram that the periodic scheduling processing time section of the embodiment of the invention is divided;
Fig. 5 a is the subdivision graph based on the real-time request queue of asking stream of the embodiment of the invention;
Fig. 5 b is the real-time subdivision graph of asking the processing time section based on request stream of the embodiment of the invention;
Fig. 6 is that the periodicity strategy based on priority carries out the flow chart that the first resource request is chosen in the scheduler of the embodiment of the invention;
Embodiment
Embodiment to concrete device of the present invention is elaborated below with reference to diagram.Dispatch control method of the present invention and device are applicable to MMNC and provide in the server of shared resource (comprising text data, multi-medium data, application program, result of calculation) visit.In following concrete grammar embodiment, the shared file resource request being carried out scheduling controlling and the shared file service is provided with server is that example describes.
Fig. 1 has described the network interconnection environment of the inventive method and device specific embodiment.It is online that the MMNC that distributes is connected in a computer local area.Also has a server that is physically located at the different location on the network.Between MMNC and the server with Client work.As shown in Figure 2, by the PPSA file request of certain format, a plurality of clients can and be sent to server and send shared resource request (request queue of client shown in the figure).Scheduling controlling device (right-hand component among the figure) works in (not shown) on the MMNC server, this scheduling controlling device comprises reception formation parts, grader, client application information table parts, comprises t0, t1, t2, the multipriority queue and the scheduler of four subqueues of t3.Server provides the file resource share service for MMNC seriatim under the scheduling controlling of scheduling controlling device.
The present invention is as follows in conjunction with the concrete dispatch control method of said apparatus:
Step 1: the All Files request at first joins the reception queue buffer (in the present embodiment by the sequencing of the time of advent, before grader is classified to file request, file request does not have type classification, having marked the type of request among Fig. 2, is that type classification should be arranged for the supporting paper request just).
For shared resource visit is dispatched effectively, the scheduling controlling device need be to the processing of classifying of the file request with different QoS requirements.Different file request is different to the requirement of server response time.On the one hand, the acceptable stand-by period of different application is different: general data are used, as Telnet, as long as its operating lag is less than or equal to 800ms, all within user's tolerance interval.And media play is used, and as MPEG-1, then requires to postpone end to end less than 250ms, shakes within 1ms end to end.On the other hand, application process is in its life cycle stage, its telefile request also is different to the ageing requirement of server response time: in the application process operation phase, skip leaf and interrupt mainly producing by calling new processing procedure or visiting new data file, only need (or several at most) file of visit in this case mostly, and the blanking time of the interruption of skipping leaf is longer, request density is little, and the delay of the request of data of this relative dispersion is little to using performance impact.But, promptly during application initializes, need visit heap file at short notice in the load phase of using, usually have tens even up to a hundred, the number of file request is more huge, if the request operating lag is bigger, the a large amount of delay adds up, and can have a strong impact on the loading performance of application.Based on above-mentioned consideration, the present invention will be divided into following 4 classes to the file request of server, and demarcate priority from high to low by t0, t1, t2, t3 order.
1. real-time files request: be generally audio frequency, video class file request.This class request can (Constant Bit Rate, CBR) class require the shake of minimum delay, minimum and fixing bandwidth to guarantee that t0 represents with symbol corresponding to fixed bit rate;
2. remote application load request: this class request can be corresponding to real-time variable bit rate (Real-time Variable BitRate, VBR) class, its generation and transmission have sudden and certain continuity, also require minimum delay, the request of priority ratio real-time files is low slightly, and t1 represents with symbol;
3. generic-document request: comprise request to system file and data file.This class request can corresponding to Non-Real Time Variable Bit Rate (non-real-time Variable Bit Rate, nrt-VBR) class has the bursty data characteristic, and allows data transfer delay to a certain degree, t2 represents with symbol;
4. pre-operation file request: be to improve the I/O of system performance, can be by analysis to historical visit information, the I/O pre-operation is carried out in following request of prediction.(Available Bit Rate, ABR) class do not have to postpone requirement, and do not require necessarily processedly, handle when idle in system, if system is busy, do not obtain handling then abandoning in certain time limit, and t3 represents with symbol corresponding to adaptive bitrate in this class request;
In order to identify different classes of file request effectively, the PPSA file request is except basic file request information, and also subsidiary have extra application message as parameter.The PPSA file request can be expressed as by client identification C_id, client and send the Process identifier Pid of request, the code Op_code of solicit operation and 4 tuples that operational parameter data district Para_buf is formed:
<C_id、Pid、Op_code、Para_buf>
Wherein, Op_code describes the operation requests to server, as GETATTR, SETATTR, LOOKUP, ACCESS, READ, WRITE, OPEN, CLOSE, MKDIR, REMOVE, RMDIR, RENAME, READDIR, FSSTAT, FSINFO, COMMIT etc., Para_buf describes and carries out certain solicit operation parameters needed.Compare with the generic-document request, comprise extra process relevant information in the parameter of PPSA request, for example:
Operation: LOOKUP parameter: (dir_fh, filename, P_comm) return value: fh
LOOKUP presses the specified path locating file on server, the handle of the file that checks in is returned client.Wherein dir_fh and filename indicate catalogue handle and the filename that will search, and P_comm is the pairing application name of process that client is sent file request.
Operation: OPEN parameter: (fh, mode) return value: flag
Open specified file, return value indicates operates successfully or fails.
Operation: READ parameter: (fh, offset, count, P_duration, P_comm) return value: (Data_buf, buf_len)
The data of file assigned address are read in the READ operation, and it is returned to client.Parameter P_duraion is the client time that process has been moved when sending file request.Return value indicates buffering area and the size thereof that data streams read is deposited.
Operation: CLOSE parameter: (fh) return value: flag
Close specified file.
Step 2: for file request is effectively classified, the information table parts of scheduling controlling device record the application message of the MMNC of current accessed server and (consider based on systematic function, only record several network applications commonly used, available profile is limited in another embodiment).The content of the application message of MMNC as shown in Figure 3, wherein, the state information of MMNC application process is recorded in data structure Client_process[n] in (we use the IP address as the MMNC device identification here, can certainly adopt other sign), the wherein corresponding MMNC in each unit; Process points to a dynamic queue, write down the network application process list of the current operation of certain client, the process of the corresponding corresponding MMNC of each node wherein, in individual process information Process_info, last_access represents the nearest access time of this process, transmission is supported if this process needs real-time files, as the media play process, then points to respective media files by rt_fh; Need the fileinfo rt_file_info that supports in real time to comprise Mean Speed rt_rate and the interior actual transfer rate cycle_rate of current dispatching cycle that needs real-time Transmission.
In PPSA, the application message of client is carried out record by scheduling controller: when file request arrives, as this process not in the process list of relative client, then this process is a new application process, watch the P_comm parameter, wish the application commonly used of writing down if this process is the user, then this progress information (pid, P_comm, last_access) is added the process tabulation of client; If server is not received the file request that certain process is new in a period of time, promptly as current_time-last_access>T
No_response, can be considered this process and finish, its deletion from tabulation; When the access request for audio or video file file arrives, the progress information that query requests should be operated, if corresponding process is the media play process, then this document transmission needs to support in real time, the rt_fh tabulation that this document is added the map network process, from the header information of this media file, read the required playback rate of this document simultaneously, insert the rt_file_info structure, simultaneously this value is joined the total real time rate of shared-file system (total_rt_rate); On the contrary, when the CLOSE request arrived, file was that media file and process are the media play process, then deletes counter structure, and deduct corresponding rt_rate from total_rt_rate.
Step 3: according to aforesaid sorting technique, the file request that comes from MMNC is divided into the real-time files request, the remote application load request, generic-document request and pre-operation file request, corresponding priority level Wei t0, t1, t2, t3. grader is the file request classification according to application message in the scheduling controller, and it is added the subqueue of corresponding priority in the multipriority queue.Grader is as follows for the calibration process of file request priority: be operating as example with Read, after file request arrives, access control apparatus is at first inquired about this and is asked corresponding client process information Process_info, if the coupling of the handle fh in the handle of request and this structure rt_fh formation in a certain rt_file_info structure, then this asks to be the request of t0 class, adds the t0 request queue; Request occurs in the load phase of remote application, i.e. P_duration<t as this document
Load(P_comm) (t wherein
Load(P_comm) for using the average load time of P_comm, can be by the configuration file setting), then this asks to be the request of t1 class, adds the t1 request queue; As historical visit information according to existing request, wish to carry out some file pre-operations, these requests are the t3 class, add the t3 request queue; Other file request all is considered as the t2 class, adds the t2 request queue.After file request adds corresponding classification queue, dispatch by the scheduler unification.In optional embodiment, demarcating for the priority of file request can be by keeping a configuration file in server, and network application and the corresponding priorities of listing support in advance are other.In another optional embodiment, also can in the additional information of file request, increase a priority level parameter, specify when initiating a file request by MMNC (being client), and to the server transmission.
Step 4:, adopted periodic scheduling algorithm PPSA that the pending file request in the multipriority request queue is dispatched in the scheduler based on priority based on the consideration of priority and fairness two aspects.Concrete grammar is as follows:
(41) file request from client adds different request queues after the grader classification, and the priority of formation reduces successively by t0, t1, t2, t3 order.File request in the high-priority queue is had relatively high expectations to ageing, wishes to obtain priority treatment, to reduce the response time.Scheduler is not promptly considered Low Priority Queuing state (as stand-by period, waiting list length etc.) if strictly follow the preferential strategy of high priority requests, as long as there is high priority requests, then handles high priority requests earlier.This strategy is under the situation of high priority requests burst, and the file request of low priority may have no chance to obtain to handle for a long time, even takes place overtime.Based on the consideration of fairness aspect, the file process among the PPSA adopts a kind of periodic scheduling strategy.Fig. 4 is the schematic diagram that the periodic scheduling processing time section of the embodiment of the invention is divided, among the figure, the time period T (can be configured according to the actual conditions of system and the performance of server) that at first is divided into one by one appropriate length the processor processing time is as dispatching cycle.PPSA gives all kinds of requests with the timeslice quota according to certain pro rate in dispatching cycle T, forms the privately owned processing time section of all kinds of requests: the time period T that handles the real-time files request
Rt, handle the time period T of load request
Load, handle general data request T
DataTime period and handle the time period T of pre-operation file request
Pre, the timeslice quota in the privately owned time period preferentially is used for corresponding class processing of request.
The processing cycle of each time period is respectively T
Rt, T
Load, T
DataAnd T
Pre, its initial value is allocated as follows:
T
pre=σ%*T
T
load=T
data=(T-T
real-T
pre)/2
In Trt, ∑ v
iRepresent all flow sums of request in real time, V is the total bandwidth of system, and α is a regulatory factor, guarantees that the bandwidth of reserving satisfies request needs in real time.In whole timeslice allocation of quota dispatching cycle process, request is in real time handled the time period of request in real time and preferentially is protected because priority is the highest.As it not being limited, when the network application process of supporting in real time when needs was too much, request in real time is occupied bandwidth or even exhaust bandwidth too much, and the request of other class can not get effective processing.For avoiding this kind situation to take place, ask shared bandwidth should have certain upper limit Trt-limit in real time, i.e. application in real time should be carried out access control.When certain network application process needs real-time service support, server at first must be inquired about current real-time traffic, if current real-time traffic summation has surpassed the service upper limit, even Trt>Trt-limit then points out the current real-time request of user to surpass the ability to bear of system.In Tpre, σ % is a proportionality coefficient, can be set by the user.
The bandwidth demand of request proposes by a plurality of network application processes are common in real time, and server not only will guarantee to ask in real time the disposed of in its entirety time T
Rt, also to satisfy the real-time requirement of single network application process, promptly to guarantee processing time T from each individual requests stream of heterogeneous networks process
Rt(i).The segmentation based on the real-time request queue of asking stream of the embodiment of the invention reaches the real-time segmentation of asking the processing time section based on request stream, shown in Fig. 5 a and Fig. 5 b, in real time request t0 can be decomposed into request stream t0 (1), t0 (2) from the different application process ... t0 (n), processing time section T in real time
RtCan be subdivided into a plurality of minor time slice T
Rt(i), T
Rt(2) ... T
Rt(n), be respectively corresponding request stream service is provided.
T
LoadAnd T
DataBe its initial value, both sums are by T
RealAnd T
PreDecision (T
Load+ T
Data=T-T
Real-T
Pre), when real-time application demand changed, this was worth also dynamic change; In addition, for satisfying the dynamic change of request of data and load request bandwidth demand, both can dynamically adjust ratio shared in dispatching cycle, in dispatching cycle, if cycle_data_rate is greater than R
Date, and cycle_load_rate is less than R
Load, then can suitably strengthen T
DataShared ratio is provided with when begin new dispatching cycle.
In PPSA, the parameter that really embodies the processing time occupation rate is the reply data byte flow that is sent after the system handles file request, represents with R here, and in period T, the corresponding flow of all kinds of requests is as follows:
R
rt=T
rt*V=∑v
i*T(1+α),R
rt(i)=v
i*T
R
load=T
load*V
R
data=T
data*V
The R here
Rt(i) the reply data byte flow of expression server to flowing from the real-time request of different application process, it is worth the rt_rate in the corresponding rt_file_info structure.With R
Rt(i), R
Load, R
DataCorresponding, the actual request data traffic in each period T uses cycle_rate (i), cycle_load_rate, cycle_data_rate to represent respectively.When begin dispatching cycle, calculate R according to the actual conditions of system
Rt(i), R
LoadAnd R
Data, simultaneously cycle_rate (i), cycle_load_rate, cycle_data_rate are put 0.
(42) all kinds of requests still have the differentiation of priority, if having the multiclass request to satisfy schedulable condition at synchronization, and each request class all has remaining timeslice, then handles real-time class request, load request and general data request successively according to priority height; Timeslice quota as certain class request in dispatching cycle uses up, even then the priority of such request is high again, also must hand to the low priority request that still has the timeslice quota to disposal right; If certain class request time sheet quota has more than neededly in dispatching cycle, remaining timeslice can be utilized by other class request, and high priority requests is preferential; After first three class request all disposes,, then carry out the pre-operation processing of request if still have the free time sheet.
(42) method of choosing pending request in is as follows:
Algorithm?Begin
Current_request=NULL;
Step1. choose request based on the timeslice quota
The real-time request queue queue of if (t0) non-NULL then
For Request (i) in queue (t0) sequential query
fh=Request(i).fh
process=Process_info(Request(i).pid);
file_handle=process.rt_fh(fh)
if?file_handle.cycle_rate+Request(i).count<
file_handle.rt_rate?then
current_request=Request(i);
file_handle.cycle_rate+=Request(i).count;
Goto?Step4;
end?if
end?if
If load request formation queue (t1) Fei Kong ﹠amp; ﹠amp; Cycle_load_rate<R
LoadThen
Select queue (t1) head request as current_request
cycle_load_rate+=current_request.count;
Goto?Step4;
end?if
If general data request queue queue (t2) Fei Kong ﹠amp; ﹠amp; Cycle_data_rate<R
DataThen
Select queue (t2) head request as current_request
cycle_data_rate+=current_request.count;
Goto?Step4;
end?if
Step2. choose based on priority
Inquire about queue (t0), queue (t1), queue (t2) successively
If formation non-NULL then
Get this formation head request as current_request, corresponding statistical value adds up
Goto?Step4;
end?if
Step3. handle the pre-operation request
If pre-operation formation non-NULL then
Get queue (t3) head request as current_request
end?if
Step4. handle request current_request
Algorithm?End
In the said method, Request describes the file request from the MMNC application process, the data length that its member variable pid, fh and count represent the identifier of the process of the request of sending, the file handle of being asked respectively and asked.
Fig. 6 carries out the idiographic flow that pending request is chosen according to the method described above, may further comprise the steps: at first choose request according to the operating position of timeslice quota, if a plurality of formations all have sheet remaining time, then choose (among the figure 601,604,606) according to priority, wherein when choosing real-time request, also must satisfy the timeslice limit (among the figure 602) of the single file request stream of using from distinct device; Secondly, if the timeslice quota of existing request queue all uses up, but the entire process cycle still have remaining timeslice, then choose request and handle (among the figure 608) according to the priority of request; At last, if after first three class request all disposes, if still have the free time sheet, then carry out pre-operation processing of request (among the figure 612).
Above-described embodiment is just carried out example to the present invention, those skilled in the art are in the technical scheme of carrying out on the basis that does not break away from the substantive thought of the present invention obtaining after various modifications and the improvement, all be considered as within the scope of the present invention, applicable scope of the present invention is as the criterion with claims and determines.
Claims (8)
1, a kind of dispatch control method of shared resource visit is characterized in that: be used for managing multi-media network computer MMNC server, may further comprise the steps:
1) receives resource request according to the sequencing of the time of advent, put into the reception formation from MMNC;
2) relevant information of network application among the record MMNC;
3) according to the relevant information of using, successively the resource request in the described reception formation is classified, it is divided into plurality of classes, and be the different priority of giving of all categories; Join the formation afterbody of corresponding priority in the multipriority request queue respectively through sorted resource request;
4) carry out the first resource request from the multipriority request queue and choose, concrete steps are:
(41) handle the cycle with the time period of a regular length as the scheduling of resource request, quantity according to priority in each dispatching cycle is divided into a plurality of processing time sections, the corresponding priority of each processing time section, each time period is according to system mode dynamic assignment regular hour sheet quota;
(42) in a scheduling processing cycle, if in the time of will from the multipriority request queue, choosing a resource request, if with there being a plurality of resource request that are positioned at different priorities request queue head to satisfy schedulable condition, and the processing time section of corresponding each priority query all also remains time enough sheet quota, then the resource request in the priority treatment high-priority queue; If the resource request of certain priority request formation head satisfies schedulable condition, but the processing time sheet quota of this priority correspondence uses up, even then its priority is high again, also must hand to the lower priority request that corresponding processing time section still has sheet quota remaining time to disposal right; In addition, if certain class request processing time section has timeslice quota more than needed, but No Assets request temporarily in the formation of corresponding priority, then remaining timeslice quota can be by the resource request utilization of all the other priority, and wherein the higher priority class resource request preferentially obtains handling.
2, dispatch control method as claimed in claim 1 is characterized in that: the resource request in the described step 1) has the relevant information of application as parameter.
3, as the dispatch control method of claim 1 or 2, it is characterized in that: the relevant information of described application comprises process identification (PID), process classification, process running environment parameter, process Visitor Logs and QoS requirement.
4, dispatch control method as claimed in claim 1 is characterized in that: described shared resource comprises text data, multi-medium data, application program and result of calculation.
5, dispatch control method as claimed in claim 1 is characterized in that: in the described step 3) resource request is divided into following classification: real-time files request, remote application load request, generic-document request, pre-operation file request, its priority reduces successively.
6, dispatch control method as claimed in claim 1 is characterized in that: described MMNC comprises computer and embedded device.
7, a kind of scheduling controlling device of shared resource visit, it is characterized in that: be used for managing multi-media network computer MMNC server, described scheduling controlling device comprises:
Receive the formation parts, be used for receiving resource request in proper order from MMNC according to the time order and function that arrives;
Client application information table parts are used for writing down the relevant information that MMNC uses;
Grader according to the relevant information of using, is classified to the resource request in the described reception formation successively, and it is divided into plurality of classes, and is the different priority of giving of all categories;
Multipriority request queue parts comprise a plurality of subqueues unit, and wherein the corresponding priority in each subqueue unit joins in the subqueue unit of corresponding priority through the sorted resource request of grader;
Scheduler is used for carrying out resource request from described multipriority request queue parts and chooses.
8, scheduling controlling device as claimed in claim 7, it is characterized in that: described MMNC comprises computer and embedded device.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200410062763 CN1237767C (en) | 2004-07-09 | 2004-07-09 | A resource access shared scheduling and controlling method and apparatus |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 200410062763 CN1237767C (en) | 2004-07-09 | 2004-07-09 | A resource access shared scheduling and controlling method and apparatus |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1606301A CN1606301A (en) | 2005-04-13 |
CN1237767C true CN1237767C (en) | 2006-01-18 |
Family
ID=34764513
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 200410062763 Expired - Fee Related CN1237767C (en) | 2004-07-09 | 2004-07-09 | A resource access shared scheduling and controlling method and apparatus |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1237767C (en) |
Families Citing this family (46)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1897566B (en) * | 2005-07-14 | 2010-06-09 | 中兴通讯股份有限公司 | System and method for realizing convergent point service quality guarantee based on class grading |
CN100445971C (en) * | 2005-11-22 | 2008-12-24 | 华为技术有限公司 | Method and device for controlling user access to shared medium |
CN100385863C (en) * | 2005-12-29 | 2008-04-30 | 华为技术有限公司 | Media resource distributing method |
KR100969768B1 (en) * | 2006-06-27 | 2010-07-13 | 삼성전자주식회사 | Apparatus and method for scheduling data in a communication system |
CN101166208B (en) * | 2006-10-20 | 2010-10-06 | 北京直真节点技术开发有限公司 | A method and system for maintaining work automation |
CN101483446B (en) * | 2008-01-08 | 2014-07-30 | 株式会社Ntt都科摩 | Method and apparatus for dynamically controlling uncontinuous receiving |
CN101290668B (en) * | 2008-06-16 | 2011-08-17 | 中国移动通信集团湖北有限公司 | Time sharing operation dynamic dispatching method and device |
CN101556563B (en) * | 2009-05-25 | 2010-10-27 | 成都市华为赛门铁克科技有限公司 | Method for controlling multi-data source access, device and storage system thereof |
CN101895544B (en) * | 2010-07-15 | 2013-10-30 | 普联技术有限公司 | Multi-user control method and multi-user control system for communication equipment and server |
EP2466852A1 (en) * | 2010-12-17 | 2012-06-20 | Swisscom AG | Digital content management |
CN102298539A (en) * | 2011-06-07 | 2011-12-28 | 华东师范大学 | Method and system for scheduling shared resources subjected to distributed parallel treatment |
US20130066943A1 (en) * | 2011-09-13 | 2013-03-14 | International Business Machines Corporation | Application-Aware Quality Of Service In Network Applications |
CN102685131A (en) * | 2012-05-10 | 2012-09-19 | 苏州阔地网络科技有限公司 | Message scheduling and controlling method and system of cloud conference |
CN102685251A (en) * | 2012-05-31 | 2012-09-19 | 苏州阔地网络科技有限公司 | Message scheduling control method and system of web conference |
CN102685252A (en) * | 2012-05-31 | 2012-09-19 | 苏州阔地网络科技有限公司 | Method and system for information control of netmeeting |
CN102685032A (en) * | 2012-05-31 | 2012-09-19 | 苏州阔地网络科技有限公司 | Method and system for processing information of net meeting |
CN102833356B (en) * | 2012-09-24 | 2015-04-29 | 河海大学 | Greedy asynchronous JavaScript and XML (AJAX) request fast response scheduling algorithm |
CN102981973B (en) * | 2012-11-05 | 2016-02-10 | 曙光信息产业(北京)有限公司 | Perform the method for request within the storage system |
CN102937915B (en) * | 2012-11-28 | 2016-02-24 | 中国人民解放军国防科学技术大学 | For hardware lock implementation method and the device of polycaryon processor |
CN103095707B (en) * | 2013-01-17 | 2015-09-09 | 苏州亿倍信息技术有限公司 | A kind of information processing method of terminal security and system |
CN103118014B (en) * | 2013-01-17 | 2015-09-09 | 苏州亿倍信息技术有限公司 | A kind of control method of terminal security and system |
CN103095706B (en) * | 2013-01-17 | 2015-09-09 | 苏州亿倍信息技术有限公司 | A kind of processing method of terminal security and system |
CN103179009B (en) * | 2013-04-22 | 2016-01-13 | 烽火通信科技股份有限公司 | A kind of dynamic self-adapting call method of distributed management system |
CN105637475B (en) * | 2014-09-16 | 2019-08-20 | 华为技术有限公司 | Concurrent access method and system |
CN104866238B (en) * | 2015-05-25 | 2018-12-14 | 华为技术有限公司 | Access request dispatching method and device |
CN106470228B (en) * | 2015-08-19 | 2020-12-15 | 研祥智能科技股份有限公司 | Network communication method and system |
CN105204928A (en) * | 2015-10-30 | 2015-12-30 | 上海斐讯数据通信技术有限公司 | Shared resource access method and system |
CN105320571B (en) * | 2015-11-27 | 2019-05-10 | 浪潮(北京)电子信息产业有限公司 | A kind of process scheduling method |
CN105490956A (en) * | 2015-12-07 | 2016-04-13 | 方正移动传媒技术(北京)有限公司 | Network request processing method and network request processing device |
CN107018091B (en) * | 2016-02-29 | 2021-04-27 | 阿里巴巴集团控股有限公司 | Resource request scheduling method and device |
CN106911592B (en) | 2016-06-01 | 2020-06-12 | 创新先进技术有限公司 | Self-adaptive resource allocation method and device |
CN106095557B (en) * | 2016-06-20 | 2019-05-28 | 浪潮(北京)电子信息产业有限公司 | A kind of improved method and device of O1 dispatching algorithm |
CN106095115B (en) * | 2016-06-30 | 2019-05-17 | 北京智能管家科技有限公司 | A kind of application program scheduling processing method and processing device |
CN106920025A (en) * | 2016-10-27 | 2017-07-04 | 蔚来汽车有限公司 | Shared resource dispatching method and system |
CN108737457B (en) * | 2017-04-14 | 2021-06-22 | 华为技术有限公司 | Method and device for processing request |
CN108093272B (en) * | 2017-12-29 | 2020-11-24 | 北京奇艺世纪科技有限公司 | Video CDN scheduling optimization method and device |
CN108833356B (en) * | 2018-05-21 | 2020-11-03 | 常州信息职业技术学院 | Multi-path queue shared resource access control device and control method |
CN109412973B (en) * | 2018-09-19 | 2022-09-13 | 咪咕数字传媒有限公司 | Audio processing method and device and storage medium |
CN110008187B (en) * | 2018-12-18 | 2023-05-12 | 创新先进技术有限公司 | File transmission scheduling method, device, equipment and computer readable storage medium |
CN109992938A (en) * | 2019-03-25 | 2019-07-09 | 大夏数据服务有限公司 | A kind of method of multi-user access computer data |
CN110855646B (en) * | 2019-10-31 | 2021-08-06 | 苏州经贸职业技术学院 | Approach loading system for network computer and application method |
CN112835508A (en) * | 2019-11-22 | 2021-05-25 | 华为技术有限公司 | Method and device for processing data |
CN111444012B (en) * | 2020-03-03 | 2023-05-30 | 中国科学院计算技术研究所 | Dynamic resource regulation and control method and system for guaranteeing delay-sensitive application delay SLO |
CN111491015B (en) * | 2020-03-31 | 2022-11-11 | 北京金山云网络技术有限公司 | Preheating task processing method and system, proxy server and service center |
CN112817765B (en) * | 2021-02-08 | 2024-06-04 | 北京同有飞骥科技股份有限公司 | Distributed object storage barrel quota management method and system based on pre-authorization |
CN115225720A (en) * | 2022-09-20 | 2022-10-21 | 亿咖通(北京)科技有限公司 | Video stream distribution method, controller, vehicle, storage medium, and program product |
-
2004
- 2004-07-09 CN CN 200410062763 patent/CN1237767C/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN1606301A (en) | 2005-04-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN1237767C (en) | A resource access shared scheduling and controlling method and apparatus | |
CN110769278B (en) | Distributed video transcoding method and system | |
CN1097229C (en) | Multimedia resource reservation system | |
CN102915254B (en) | task management method and device | |
US9391857B2 (en) | Scheduling requests for data transfers in a multi-device storage system | |
US6725456B1 (en) | Methods and apparatus for ensuring quality of service in an operating system | |
US11586392B2 (en) | Multi-stream SSD QoS management | |
US20080109452A1 (en) | Queuing model for a plurality of servers | |
KR20050020942A (en) | Continuous media priority aware storage scheduler | |
CN1610347A (en) | Methods, apparatus for managing performance and resource utilization within cluster-based systems | |
US20170171337A1 (en) | Scalable data feed system | |
US9817698B2 (en) | Scheduling execution requests to allow partial results | |
CN1921449A (en) | Stable, minimal skew resource flow control method and system | |
CN103139100A (en) | Business handling method and business handling system | |
US20140108458A1 (en) | Network filesystem asynchronous i/o scheduling | |
CN113132456B (en) | Edge cloud cooperative task scheduling method and system based on deadline perception | |
CN110708256A (en) | CDN scheduling method, device, network equipment and storage medium | |
US20040221011A1 (en) | High volume electronic mail processing systems and methods having remote transmission capability | |
CN107040475B (en) | Resource scheduling method and device | |
CN116383240A (en) | Multi-database acceleration query method, device, equipment and medium based on FPGA | |
Mokbel et al. | Scalable multimedia disk scheduling | |
Balafoutis et al. | Clustered scheduling algorithms for mixed-media disk workloads in a multimedia server | |
Nerjes et al. | Incremental scheduling of mixed workloads in multimedia information servers | |
CN107172142B (en) | A kind of data dispatching method accelerating cloud computation data center inquiry | |
EP1753237B1 (en) | Scheduling for Internet Protocol Television Broadcast |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20060118 Termination date: 20180709 |