CN1271519C - Self-regulating dispatching method of streammedia server resources - Google Patents
Self-regulating dispatching method of streammedia server resources Download PDFInfo
- Publication number
- CN1271519C CN1271519C CN 200410027921 CN200410027921A CN1271519C CN 1271519 C CN1271519 C CN 1271519C CN 200410027921 CN200410027921 CN 200410027921 CN 200410027921 A CN200410027921 A CN 200410027921A CN 1271519 C CN1271519 C CN 1271519C
- Authority
- CN
- China
- Prior art keywords
- task
- streaming media
- dispatching
- activation condition
- media
- 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
Images
Landscapes
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
- Information Transfer Between Computers (AREA)
Abstract
The present invention relates to a self-regulation dispatching method of streaming media server resources, which includes the following steps: the position of each task of a steaming media task set in a ready queue is determined by an RM algorithm, and activation conditions are set for tasks with low priority; a despatcher dispatches and sends media data according to the ready queue and monitors image quality; if the image quality does not meet requirements, the despatcher monitors whether CPU idle time exists or not between the moment when the current task is finished and the moment when the next task begins, if true, executing the task i whose image quality is affected, else, checking whether the activation conditions of the task i meet the requirements, if true, changing the dispatching parameters of the task i and adjusting the data queue of streaming media for dispatching the media data again, else, executing the task i according to the original dispatching sequence. The priority level of the tasks can be modified in real time according to the quality of steaming media serve and resource utilization states, and the service quality of the steaming media and the utilization rate of a server are improved.
Description
Technical field
The present invention relates to stream media technology, particularly a kind of self-regulation dispatching method of streaming media server resource, the priority that it comes the real time modifying task according to the quality and the utilization of resources situation of streaming media service, thereby the resource utilization of raising streaming media service quality and server.
Background technology
Along with further popularizing of internet, the particularly development of broadband network technology, people are to the increase day by day of multimedia messages demand, and it is a kind of that new the multimedia wideband network technology---stream media technology is arisen at the historic moment.Stream media technology is a comprehensive notion technology, and it comprises Video processing (video compress, decompress(ion), coding, decoding etc.) technology, network transmission technology, dispatching technique, distributed proccessing or the like.Streaming Media refers to use the continuous time-base media of stream transmission technology in network, as: audio frequency, video or multimedia file.Streaming Media is not downloaded whole file before broadcast, only will begin partial content deposits internal memory in, and the data stream of Streaming Media transmits at any time at any time and plays, and some delays are just arranged when beginning.
Streaming media service is meant that streaming media server uses stream media technology to provide desired service by network to the client who sends the Streaming Media demand.Streaming media service has boundless market application foreground, and its application comprises multimedia news briefing, online live, network video advertisement, ecommerce, video request program, long-distance education, network radio station, real-time video meeting and emerging interactive game, three-dimensional animation, interactive training etc.The quality of streaming media service mainly is subjected to network, server and client side's Effect on Performance.Along with the development of computer hardware technique, the particularly reinforcement greatly of CPU processing power, client performance no longer has been the subject matter that influences streaming media service quality.Under the certain situation of network condition, streaming media server has played conclusive effect to service quality.And convection media server, dispatching algorithm is its core, and the resource that good dispatching algorithm can effectively utilize server satisfies a large amount of services request, and the load of balance server prevents that effectively server is because of service overload " death ".
Streaming media service has certain concurrency, may exist a plurality of users to ask streaming media service in the identical moment, the streaming service has the certain time sequence requirement, and some task must be finished in certain time limit, so the dispatching algorithm of streaming media server all adopts the real-time system dispatching algorithm.The real-time scheduling realizes scheduling to real-time task by the resource of reasonably arranging real-time server, promptly how Resources allocation produces correct output to guarantee each real-time task in the correct period, and the uncertainty that task is exported is limited in certain scope.The resource of so-called server mainly is the finger processor time resource, reasonably distributes the operation order of the time resource task that is exactly that reasonably the arrangements are complete.Real-Time Scheduling is divided into two stages, and the phase one is to determine the position of each ready task in ready queue, and subordinate phase is when task is switched, and makes first task in the ready queue obtain processor.In real-time system, the real-time scheduling decides at a time call which real-time task according to the priority of real-time task.
Existing streaming media server resource regulating method derives from traditional real-time scheduling, commonly used is a kind of typical static priority dispatching algorithm-RM (Rate-Monotonic) dispatching algorithm, the RM dispatching algorithm decides dispatching priority according to the length in task executions cycle, the task executions cycle is longer, its dispatching priority is lower, and the priority of task remains unchanged in scheduling process.
The RM dispatching algorithm is that standard judges whether this task-set can dispatch, and the model of its real-time task collection can be expressed as follows with the resource utilization of real-time task collection: (τ 1, and τ 2 for real-time task collection S, ..., τ n), task is arranged to increase preface by the cycle size, T1≤T2≤...≤Tn.The RM algorithm is by the priority of the length in task executions cycle definition task, and the shortest task priority of performance period is the highest, and the longest task priority of performance period is minimum.Task priority is arranged as follows among the real-time task collection S: P1 〉=P2 〉=.. 〉=Pn.Convection current media task collection S RM algorithmic dispatching according to above-mentioned task model, supposes that the priority of task among the task-set S2 always will be lower than the priority of task among the task-set S1, and if have its LT of a task
1i≤ T
2k＜L (T
1i+ C
1i) (wherein L is a natural number, T
1i, C
1iBe respectively the cycle and the execution time of i task among the task-set S1, T
2kBe the cycle of k task among the task-set S2), then this task will can not be performed less than cpu resource because of competition, and this is because in the moment that this task each time arrives, CPU always can not find time carry out this task in other task of execution.And in Streaming Media was used, this periodic task was ubiquitous.Because the stream medium data amount is often very big, the resource contention fierceness, adopt traditional RM dispatching algorithm to tend to cause a part of media task because competition can not get less than the resource of server carrying out, thereby produce the quality problems of streaming media service, as image blurring, shake, overlapping, jump, pause etc.On the other hand, the RM dispatching algorithm is the static scheduling algorithm, and in order to guarantee the schedulability of task, the resource that a part is valuable always can not get effectively utilizing and is wasted.
Summary of the invention
The object of the present invention is to provide a kind of RM dispatching method that can carry out self-control according to the utilization of resources situation of streaming media service quality requirement and streaming media server, to reduce the quantity that the Streaming Media task is not performed less than resource because of competition, improve the quality of streaming media service, improve the resource utilization of streaming media server.
The present invention realizes that by the following technical solutions it may further comprise the steps:
A, determine the position of each task in ready queue in the Streaming Media task-set, and set an activation condition for each low priority task with the RM algorithm;
Media data is dispatched, sent to b, scheduler by this ready queue, and monitoring picture quality;
C, if picture quality does not meet the demands, whether the monitoring current task is carved into next task when executing cpu idle time between the zero hour, if having, utilize the affected task i of this cpu idle time carries out image quality; If there is not cpu idle time, check then whether the activation condition of described task i satisfies, activation condition satisfies the scheduling parameter that then changes task i, adjusts the formation of matchmaker's fluid data, reschedules; Activation condition does not satisfy then to be carried out by former dispatching sequence.
In above-mentioned steps a, described activation condition is the threshold value of the number of tasks of losing continuously before this task, and/or this task stand-by period threshold value.
In above-mentioned steps b, the criterion of described picture quality is that image is whether fuzzy, at least a situation in shake, overlapping, jump and the pause etc., and basis for estimation is total number of executions of the task that is scheduled.
The present invention has carried out important improvement to traditional RM dispatching method, no longer the priority and the cycle of task are bound, but come the priority of real time modifying task according to the utilization of resources situation of the quality of streaming media service and server, reduced the quantity that the Streaming Media task is not performed less than resource because of competition, improved the quality of streaming media service, simultaneously, solve the defective of traditional RM dispatching algorithm waste cpu resource, improved the resource contention situation.In addition, this dispatching method calculated amount is little, and the complexity of algorithm is the same with former RM dispatching algorithm.
Description of drawings
Fig. 1 is a main flow chart of the present invention.
Embodiment
Below in conjunction with accompanying drawing the present invention is elaborated.
Traditional RM dispatching method is from the Real-Time Scheduling field, the priority that this method set the tasks according to the task executions cycle, and calculated amount is few, and is simple, and easily row is considered to the optimal static dispatching algorithm.But on streaming media server, use this method can produce some problems,, soft real-time task is arranged, also comprised the part un-real time job because Streaming Media not only has the hard real-time task in using.The multitask collection situation of this complexity of stream media system causes using traditional RM dispatching algorithm when dispatching, and the scheduling performance of system does not reach the due impact of performance in traditional real-time domain.In addition, the demand of Streaming Media and data volume are often very huge, and be very big at task amount, under the heavy situation of scheduler task, traditional RM dispatching method tends to cause the resource behaviour in service to worsen, and priority is lower but belong to critical task competition and can not carry out less than cpu resource.
Cause part Streaming Media task to be because of the main cause that competition can not be performed less than cpu resource: traditional RM algorithm is to dispatch by the priority of task, task priority then is by the size definition of duty cycle, the shortest task priority of cycle is the highest, the longest task priority of cycle is minimum, and the priority of task remains unchanged in scheduling process.That is to say that the RM dispatching method is bound the cycle and the priority of task, and do not consider the actual conditions of server resource.But in Streaming Media was used, there be not inevitable absolute relation in the cycle of Streaming Media task and the priority of this task.Can see a simple example: when the playing stream media video file, between each I frame of video flowing two P frames are arranged, it is the duty cycle that the duty cycle of I frame is greater than the P frame, but the broadcast of losing the streaming media video file of well-known P frame task can not produce too much influence, can have influence on the animation effect of video playback if I frame task is lost.But in classical RM algorithm, thus because I frame task priority will be than easier the losing of P frame task than the low I frame of P frame task priority task, and in fact relatively P frame I frame be the mission critical of video flowing.
In order to realize self-regulating RM dispatching method, the present invention has at first broken the way with duty cycle and priority binding, has proposed a kind ofly can come the self-regulation dispatching method of real time modifying task priority according to the utilization of resources situation of streaming media service quality requirement and streaming media server.
The present invention adopts traditional RM algorithm to dispatch earlier to the Streaming Media task of being asked, if satisfying quality of service requirement, scheduling result directly sends operation result, if the traditional scheduler strategy can not satisfy quality of service requirement, the strategy change scheduling parameter that then adopts the present invention to propose is dispatched again.
Its detailed process is as follows:
As shown in Figure 1, at first determine the position of each task in ready queue in the Streaming Media task-set, and set an activation condition for each low priority task with the RM algorithm.Described activation condition is the threshold value of the number of tasks of losing continuously before this task, and/or this task stand-by period threshold value.Activation condition can be given according to actual conditions, such as concerning the I frame, can define activation condition and be and allow the number of tasks (as being 2) of losing continuously before the I frame task, if the number of tasks of actual loss more than or equal to the number of tasks that allows to lose continuously, then activation condition satisfies.The threshold value that also can set the task stand-by period for the long period task is an activation condition, and activation condition satisfies when the actual stand-by period of task surpasses this threshold value.
Scheduler is dispatched, is sent the client of media data to this streaming media service of request by network by above-mentioned ready queue, and the quantity of " dying of hunger " and losing according to task is monitored picture quality simultaneously.Wherein the criterion of picture quality be that image is whether fuzzy, at least a situation in shake, overlapping, jump and the pause etc., basis for estimation is total number of executions of the task that is scheduled.
When picture quality did not meet the demands, whether the monitoring current task is carved into next task when executing cpu idle time between the zero hour, if having, utilize the affected task i of this cpu idle time carries out image quality; If there is not cpu idle time, check then whether the activation condition of described task i satisfies, activation condition satisfies the scheduling parameter that then changes task i, improves the priority of task i, adjusts the formation of matchmaker's fluid data, reschedules; Activation condition does not satisfy then to be carried out by former dispatching sequence.
In said method, the priority of Streaming Media task no longer is to link directly changeless with the cycle in the task scheduling process, but can be according to given activation condition, be quantity and the task stand-by period that task can not be competed cpu resource, change.So, the lower task of beginning part priority ratio, after certain stand-by period of process (waiting time can not surpass the time limit that task is carried out), its priority can improve, thereby in next step resource contention, take advantage, the possibility that obtains resource has increased, and the possibility that obtains carrying out has also just increased.
Be of the present invention one concrete application examples below:
1, embodiment environment
The hardware configuration of server end and operating system:
CPU：Intel?PIII?1GHz
Internal memory: 128M
Network interface card: 10M/100M self-adaptation
Operating system: RedHat7.1 Linux Server
The configuration of client and operating system:
CPU：Intel?PIII?800MHz
Internal memory: 128M
Network interface card: 10M/100M self-adaptation
Operating system: Windows2000 Professional
Software environment:
Employed server of the server and client side of embodiment and playout software are the Linux streaming media video server system software of independent development, in order to strengthen the influence that comparative effectiveness reduces other factors, client and network environment should be consistent, and network should be in good unimpeded state, can drop to Network Transmission minimum to the influence of quality like this.The server film source is but that range of code rates is 110Kpbs-1100Kpbs based on the fluidisation video file of MPEG4 coding.
2, operation result
Embodiment mode with video request program in LAN environment is moved the scheduling scheme of the present invention's proposition and traditional RM scheduling scheme respectively, monitors the utilization of resources situation of media server, task scheduling situation, the situation of number of tasks several respects of scheduling.
Under the less situation of low code check and program request client number, when being lower than 500Kpbs and program request number and being less than 5 as code check, because resource is relatively more rich, two kinds of dispatching methods almost do not have difference, can both obtain good dispatching effect.The resource of server is in the stable state that utilizes, and does not find to carry out because of serious resource contention causes task to be lost and refuse.
Along with the rising of code check and the increase of program request client number, it is more and more busier that CPU becomes, the resource contention aggravation, and advantage at this time of the present invention begins to reveal.Because traditional RM dispatching algorithm can not be regulated according to the actual operating position of resource, resource contention more and more worsens, utilization of resources instability, and the part scheduler task can not enter ready queue, and the task that part enters ready queue can not get resource.Dispatch with the method that the present invention proposes, find that then whole scheduling process resource contention is orderly, resource utilization remains on high-caliber state always, has only the small part task selectively to be lost.Under worse simulated experiment condition, with traditional RM scheduling, just have 98--100 task to be lost less than resource in per 500 stream medium datas, and when dispatching with method of the present invention, these data have dropped to below 10 because of competition.
The present invention makes amendment to the RM dispatching method, priority to task is made amendment, no longer the priority and the cycle of task are bound, but come real time modifying priority with the mass parameter of streaming media service and the utilization of resources situation of server, thereby realized the purpose of self-control, improve service quality, improved the resource contention situation.Use the RM dispatching method convection current media task of the self-control priority that the present invention puts forward to dispatch, compare, on the resource utilization of the number of executions of task and media server, very big advantage is arranged with traditional RM dispatching method.
Claims (2)
1, a kind of self-regulation dispatching method of streaming media server resource is characterized in that may further comprise the steps:
A, determine the position of each task in ready queue in the Streaming Media task-set with the dull algorithm of speed, and set an activation condition for each low priority task, described activation condition is threshold value and/or this task stand-by period threshold value of the number of tasks of losing continuously before this task;
Media data is dispatched, sent to b, scheduler by this ready queue, and monitor picture quality according to the criterion and the basis for estimation of picture quality;
C, if picture quality does not meet the demands, whether the monitoring current task is carved into next task when executing cpu idle time between the zero hour, if having, utilize the affected task i of this cpu idle time carries out image quality; If there is not cpu idle time, check then whether the activation condition of described task i satisfies, activation condition satisfies the scheduling parameter that then changes task i, adjusts the formation of matchmaker's fluid data, reschedules; Activation condition does not satisfy then to be carried out by former dispatching sequence.
2, the self-regulation dispatching method of streaming media server resource according to claim 1, it is characterized in that: among the above-mentioned steps b, the criterion of described picture quality is that image is whether fuzzy, shake, overlapping, at least a situation in jumping and pausing, and basis for estimation is total number of executions of the task that is scheduled.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200410027921 CN1271519C (en) | 2004-06-30 | 2004-06-30 | Self-regulating dispatching method of streammedia server resources |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200410027921 CN1271519C (en) | 2004-06-30 | 2004-06-30 | Self-regulating dispatching method of streammedia server resources |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1595362A CN1595362A (en) | 2005-03-16 |
| CN1271519C true CN1271519C (en) | 2006-08-23 |
Family
ID=34664078
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200410027921 Expired - Fee Related CN1271519C (en) | 2004-06-30 | 2004-06-30 | Self-regulating dispatching method of streammedia server resources |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1271519C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101448006B (en) * | 2008-12-25 | 2012-01-11 | 中兴通讯股份有限公司 | Method and system for realizing access for a great amount of terminals for streaming media server |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100423504C (en) * | 2005-05-12 | 2008-10-01 | 复旦大学 | Big cocurrent media real-time transmission method |
| CN101370308B (en) * | 2005-05-26 | 2011-03-30 | 华为技术有限公司 | Method for implementing channel priority level scheduling and business priority scheduling |
| US7664856B2 (en) * | 2005-07-28 | 2010-02-16 | Microsoft Corporation | Dynamically balancing user experiences in a multi-user computing system |
| US8458711B2 (en) * | 2006-09-25 | 2013-06-04 | Intel Corporation | Quality of service implementation for platform resources |
| KR101342365B1 (en) | 2006-12-07 | 2013-12-16 | 엘지전자 주식회사 | Method of transferring data in wireless communication system |
| US8797879B2 (en) | 2006-12-07 | 2014-08-05 | Lg Electronics Inc. | Method of transmitting and receiving status report in a mobile communication system |
| JP5113186B2 (en) | 2006-12-07 | 2013-01-09 | エルジー エレクトロニクス インコーポレイティド | Data transmission method in wireless communication system |
| WO2008084984A2 (en) | 2007-01-09 | 2008-07-17 | Lg Electronics Inc. | Method of controlling data retransmission in a wireless communication system |
| US8155069B2 (en) | 2007-01-09 | 2012-04-10 | Lg Electronics Inc. | Method of transmitting and receiving scheduling information in a wireless communication system |
| WO2008084985A2 (en) | 2007-01-09 | 2008-07-17 | Lg Electronics Inc. | Method of transmitting and receiving data in a wireless communication system |
| KR101211758B1 (en) | 2007-01-10 | 2012-12-12 | 엘지전자 주식회사 | Method for generating block data in wireless communication system |
| KR101155237B1 (en) | 2007-01-10 | 2012-06-13 | 엘지전자 주식회사 | Method for network's transmitting data in mobile communication |
| CN101330472B (en) * | 2008-07-28 | 2013-01-16 | 中兴通讯股份有限公司 | Method for caching and processing stream medium data |
| CN114780253B (en) * | 2022-06-15 | 2022-08-30 | 广东省信息工程有限公司 | Big data intelligent analysis system |
-
2004
- 2004-06-30 CN CN 200410027921 patent/CN1271519C/en not_active Expired - Fee Related
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101448006B (en) * | 2008-12-25 | 2012-01-11 | 中兴通讯股份有限公司 | Method and system for realizing access for a great amount of terminals for streaming media server |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1595362A (en) | 2005-03-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1271519C (en) | Self-regulating dispatching method of streammedia server resources | |
| CN110769278B (en) | Distributed video transcoding method and system | |
| Ashraf et al. | Stream-based admission control and scheduling for video transcoding in cloud computing | |
| EP2256735A1 (en) | Rendering in a multi-user video editing system | |
| CN103838779A (en) | Idle computing resource multiplexing type cloud transcoding method and system and distributed file device | |
| CN111782355B (en) | Cloud computing task scheduling method and system based on mixed load | |
| CN104104973B (en) | A kind of group's Bandwidth Management optimization method for being applied to cloud media system | |
| CN111294647B (en) | Video processing method, device and equipment and storage medium | |
| GB2533017A (en) | A method and system for scalable job processing | |
| EP0682833A1 (en) | Flow control by evaluating network load. | |
| Bradshaw et al. | Periodic broadcast and patching services: implementation, measurement, and analysis in an internet streaming video testbed | |
| WO2013137987A1 (en) | Adaptive media delivery | |
| Isovic et al. | Quality aware MPEG-2 stream adaptation in resource constrained systems | |
| CN113630438A (en) | Stream processing task scheduling method and distributed stream processing system | |
| Baiceanu et al. | Multimedia applications require adaptive CPU scheduling | |
| CN109388501B (en) | Communication matching method, device, equipment and medium based on face recognition request | |
| CN113328879B (en) | Cloud data center network QoS (quality of service) guaranteeing method based on network calculus | |
| CN111199316A (en) | Cloud and mist collaborative computing power grid scheduling method based on execution time evaluation | |
| CN113742009B (en) | Desktop cloud environment resource scheduling method, device, equipment and storage medium | |
| Basu et al. | Implementation of dynamic service aggregation for interactive video delivery | |
| Poellabauer et al. | Coordinated cpu and event scheduling for distributed multimedia applications | |
| Zhao et al. | Efficient adaptive media scaling and streaming of layered multimedia in heterogeneous environment | |
| Poon et al. | Determination of efficient transmission scheme for video-on-demand (VoD) services | |
| CN109343940A (en) | Multimedia Task method for optimizing scheduling in a kind of cloud platform | |
| CN112543373A (en) | Transcoding control method and device and electronic equipment |
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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20060823 |