KR100235844B1 - Large scale information on demand system and information on demand providing service method using dynamic load balancing method - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

The present invention relates to a large-scale on-demand information providing system and method using a dynamic load balancing technique.

2. The technical problem to be solved by the invention

The present invention provides a large-scale on-demand information providing system and method using dynamic load balancing in order to efficiently use the stream processors responsible for transmission in configuring a system for a large-scale on-demand information service, such as on-demand video service. To do so.

3. Summary of Solution to Invention

The present invention adds various factors to be considered in the quality of service (QoS) and session management to the proposed EGM, expands the application, and expresses it as a normalized value. It can satisfy the real-time in providing multimedia services by applying dynamic load balancing technique, and fully improve the performance of the stream processor by applying the ease of server development, fast response speed, and effective quality of service (QoS) control technology. It can be utilized.

4. Important uses of the invention

The present invention is used in the on-demand information providing service.

Description

Large on-demand information provision system and dynamic on-demand service provision method using dynamic load balancing

The present invention provides a large on-demand information providing system and an on-demand information providing service using a dynamic load balancing technique in order to efficiently use stream processors responsible for transmission in configuring a system for a large on-demand information providing service such as an on-demand video service. It is about a method.

In the case of a conventional video system and a system server, it is possible to fully utilize the processing power of a plurality of stream processors by applying a load balancing technique using only static criteria or not applying a load balancing technique to a stream processor having a function of transmitting a stream. There was a problem that can not.

In order to solve the above problems, the present invention adds various elements that should be considered in quality of service (QoS) and session management to the existing Extended Gradient Model (EGM). It is suitable for the same on-demand information providing service and is expressed as a normalized value, thereby making full use of the service capability of the stream server constituting the server for the on-demand information providing service to provide better query service and to accommodate more users. Dynamic load balancing technique is applied to on-demand video system server for large-scale on-demand information service, satisfying real-time in providing multimedia service, ease of server development, fast response speed, and effective quality of service (QoS) control. Technology To provide on-demand information available to fully utilize the performance of the system using the stream processors and application specific information service method has its purpose.

1 is a block diagram of a large video-on-demand system according to the present invention;

2 is a configuration diagram of a system server for an on-demand video service according to the present invention;

3 is a flowchart of a large on-demand video service method using a dynamic load balancing technique according to the present invention;

4 is a detailed flowchart of a process of allocating a stream processor in FIG. 3 according to the present invention to start a service.

* Explanation of symbols for the main parts of the drawings

11,21: video stream pool 12,22: network

13,23: client

In order to achieve the above object, the apparatus of the present invention establishes a session for an arbitrary service, and is connected through a network with a plurality of other video stream pools to load balance the stream pools to handle a user's request for a service. Stream pools; At least one stream processor that receives a service from a stream pool including itself among the plurality of stream pools and transmits stream data to a user; It accepts a service request of a user, manages sessions and resources of the stream pool including itself among the plurality of stream pools, allocates a service to the stream processor managed by the user, and performs normalization operation for load balancing. Dynamic load balancing forming means; And stream data storage means for storing stream data that is striped and simultaneously read by the plurality of stream processors and transmitted to each user terminal.

In addition, the method of the present invention, the on-demand information providing service method applied to the on-demand information providing system, the stream pool receiving a service request from the user comprises a first step of determining whether the requested stream data exists in its database; A second step of transferring the service to the stream pool having the requested stream data if the requested stream data does not exist in its database as a result of the determination of the first step; And a third step of allocating the serviceable stream processor by applying the dynamic load balancing module if the requested stream data exists in its database as a result of the determination of the first step.

Hereinafter, with reference to the accompanying drawings will be described an embodiment according to the present invention;

Definitions for normalization used in the present invention are as follows.

A large on-demand video system consists of n video stream pools and m stream processors.

Video stream pool (i) represents the i-th video stream pool in the system. The stream processor (i, j) represents the j th stream processor belonging to the video stream pool (i), and the dynamic load balancing module (i) is a subject of dynamic load balancing operation in the video stream pool (i). A node that is visible to the outside world in a video system.

In the parameter used in the present invention, there is a parameter for load balancing between the internal processors of the video stream pool (i) and the stream processor (i, j) and a parameter for load balancing of the entire system. Next, the parameters for load control added or modified in the EGM are defined as follows.

First, the network overhead C (i, j) is a value considered as the communication overhead between the video stream pool (i) and the video stream pool (j) existing in the on-demand video system. These values can be changed depending on the status and network configuration.

Proximity PXt (i) is the adjacency of video stream pool (i) at time t. If video stream pool (i) is a light load, PXt (i) is 1, otherwise PXt (i) is 1 + min {PXt (j)} + min {C (i, j)}. Where j represents any video stream pool adjacent to video stream pool (i).

Proximity PXt (ij) is the adjacency of stream processor (i, j) at time t, if PXt (i, j) is 1 if video stream pool (i) is a small amount of LIGHT LOADED, In other cases, PXt (i, j) is 1 + min {PXt (i, k)}. In this case, the k th stream processor is included in the video stream pool (i), and k and j are not equal (k ≠ j).

The indentation pressure PIPt (i) is the indentation pressure of the video stream pool (i) at time t, which is [{Dt (i)-ψt (i)} x C / PXt (i)]. Where Dt is the global dynamic threshold of the on-demand video system, ψ t (i) is the load of the video stream pool i at time t, and C is a constant value on communication.

Inlet pressure PIPt (i, j) is the inlet pressure of stream processor (i, j) at time t, which is [{Dt (i)-ψt (i, j)} x β / PXt (i, j)]. Here, β is a constant value given according to the topology of the network and is the same in the video stream pool.

Dynamic Threshold (Dt) is an indicator for determining the load status of each processor.Dynamic Threshod Dt between video stream pools across an on-demand video system is D _t-1 + δ (ψt) / n. Where ψt is the number of sessions being processed throughout the video-on-demand system at time t, and the dynamic threshold Dt (i) between the stream processors of the video stream pool (i) is D _t-1 (i) + δ (ψt (i)) / n.

The Workload Distribution Surface (WDS) is a relative value that indicates the current load status of each processor, and the surface value St (i) at time t of the video stream pool (i) is-(Dt-ψt (i) / PXt (i) and WDSt is [St (1), St (2), ..., St (n)].

Surface value St (i, j) at time t of stream processor (i, j) is-(Dt (i) -ψt (i, j) / PXt (i, j), and WDSt is [St (i, 1), St (i, 2), ..., St (i, m)].

1 is a block diagram of a large-scale on-demand video system according to the present invention, which is composed of n video stream pools, and each video stream pool is composed of at most m stream processors. 2 is a block diagram of a system server for an on-demand video service according to the present invention. In the figure, numerals 11 and 21 denote video stream pools, numerals 12 and 22 denote networks, and numerals 13 and 23 denote clients.

The system for on-demand video system services with load balancing is composed of multiple video stream pools, each of which is an outstanding node, which is a video stream pool M and a stream processor. It consists of a group and has a database that stores stream data.

Each component constituting the video-on-demand system and its functions are as follows.

The video stream pool 11 is a basic unit that can constitute a session of an on-demand video system service, and processes a request for a service of a user. The video stream pool 11 can be seen as one node in a large on-demand video system, and any video stream pool 11 can be connected to another video stream pool via a network 12, and mutually share information. By doing so, load balancing on the video stream pool is achieved.

The dynamic load balancing module is a node that is exposed to the outside, and receives a user's service request, manages sessions and resources of the video stream pool 11 included therein, and allocates services to stream processors managed by the dynamic load balancing module. The operation is performed to configure WDS for load balancing.

The stream processor is obliged to send stream data to the user for the service assigned from the video stream pool 11. The stream processors constituting an arbitrary video stream pool 11 may be the same kind of processors, and the stream processors constituting different video stream pools 11 may be different models.

The stream database stores video streams such as Moving Pictures Expert Group I (MPEG I) or MPEG II, and stores resources that are read by a stream processor and transmitted to a user terminal. Stream data is striped so that it can be read by multiple stream processors at the same time.

3 is a flowchart illustrating a large on-demand video service method using a dynamic load balancing technique according to the present invention.

First, the user requests a service for stream i from video stream pool M on the video stream pool to which he is subscribed (31). At this time, the user transmits his specification and information on the title of the movie to receive the service.

The video stream pool receiving the service determines whether the requested stream data exists in its stream database (32), and if the requested stream data does not exist in its database, the video stream pool having the corresponding stream data is placed on demand video. It locates in the service system and connects the user (33). At this time, transparency is provided to the user.

On the other hand, the video stream pool checks whether a serviceable stream processor exists if the requested stream data exists in its database (34), and if the serviceable stream processor exists, allocates the service to the serviceable stream processor to provide the service to the user. (37).

On the other hand, if there is no serviceable stream processor, the dynamic load balancing module moves the load by applying a load balancing technique in the video stream pool for load balancing (35).

As a result, load balancing may be performed in the video stream pool, but the service may not be provided due to the excess capacity of the stream processors. Therefore, it is determined whether a serviceable stream processor exists (36). In operation 37, the service is provided to the user by allocating the service to the serviceable stream processor.

On the other hand, if there is no serviceable stream processor, the dynamic load balancing module moves the load by applying a load balancing technique among the video stream pools for load balancing (38), and then allocates the service to the serviceable stream processor to service the user. (37). At this time, a dedicated communication path is established between the user and the serviceable stream processor.

4 is a flowchart according to the present invention, illustrating a process of allocating a stream processor to start a service according to the dynamic load balancing scheme of FIG. That is, the dynamic load balancing module represents a process of allocating a stream processor and starting a service by performing a normalization operation process and a load movement process for applying a load balancing technique to a user's service request.

In the present invention, the following matters are assumed for performing the function of the dynamic load balancing module.

First, the dynamic load balancing module has load information for its stream processors.

Second, it contains information on the types and specifications of streams that can be serviced by its own stream processor, quality of service (QoS) as well as the streams that other stream processors in the on-demand video system have.

Third, you can request and create the information needed to configure the system-wide WDS, as well as your own workload distribution surface (WDS).

When a user requests a service for stream (j) from a stream processor (i) to which he or she is registered, a process of allocating a service received by the dynamic load balancing module (i) to an arbitrary stream processor will be described. Same as

First, when stream j is received by the dynamic load balancing module, it is checked whether it can service while satisfying the quality of service (QoS) conditions of stream (j) (41).

As an element of quality of service (QoS) on the video stream pool, identify the MPEG and encoding format, bit rate, required bandwidth, transmission rate, and dead line conditions.

Then, it is checked whether there is a serviceable stream processor among its stream processors while satisfying the quality of service (QoS) condition of stream (j) while having IDLE or LIGHT LOADED (42). If a serviceable stream processor is present, the service for stream (j) is allocated (43) to the serviceable stream processor to start the service (44). IDLE and LIGHT LOADED indicate the state of each processor and are mentioned in the EGM.

On the other hand, if there is no serviceable stream processor, it forms a WDS (i) for the video stream pool (i) and moves the load inside the video stream pool (i) (45). That is, the dynamic load balancing module obtains normalized values for load movement and moves the load in its video stream pool. Thereafter, it is determined again whether there is a serviceable stream processor (46). If a serviceable stream processor exists, the service for stream (j) is allocated to the serviceable stream processor (43) and service is started (44).

On the other hand, if there is no serviceable stream processor, it forms a WDS for the entire video-on-demand system for load balancing in the video stream pool and moves loads between the video stream pool (47). That is, the dynamic load balancing module moves the load by obtaining normalized values to move the load between the video stream pools. The service is then allocated to the stream processor for stream (j) (43) to initiate service (44).

If there is more than one video stream pool, the one with the smaller C (i, j) value is chosen. On the other hand, if the serviceable video stream pool (k) is not found among the system-wide video stream pools, the service cannot be provided because the video system on demand cannot be serviced.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the spirit of the present invention for those skilled in the art to which the present invention pertains, and the above-described embodiments and accompanying It is not limited to the drawing.

The present invention as described above, by applying a dynamic load balancing technique for large on-demand video service, various effects can be obtained as follows.

First, real-time requirements of on-demand video services can be guaranteed.

Second, quality of service (QoS) can be guaranteed for user needs.

Third, it is possible to maximize the utilization of all the processors constituting the video on demand system.

Fourth, it can be applied to various multimedia services as well as video system on demand.

Claims (7)

A stream pool for establishing a session for an arbitrary service and being connected through a network with a plurality of other video stream pools to balance load between the stream pools to process a request for a service of a user; At least one stream processor that receives a service from a stream pool including itself among the plurality of stream pools and transmits stream data to a user; It accepts a service request of a user, manages sessions and resources of the stream pool including itself among the plurality of stream pools, allocates a service to the stream processor managed by the user, and performs normalization operation for load balancing. Dynamic load balancing forming means; And And stream data storage means for storing stream data that is striped and simultaneously read by the plurality of stream processors and transmitted to each user terminal. The method of claim 1, Dynamic load balancing forming means for performing a normalization operation for the load balancing, Network overhead considered as communication overhead between the stream pools present in the on-demand information providing system, proximity of the stream pool at any time, indentation pressure of the stream pool at any time, and each processor And a formula such as a dynamic threshold for determining a load state of the processor and a workload distribution surface (WDS) indicating a current load state of each processor. In the on-demand information providing service method applied to the on-demand information providing system, The stream pool receiving the service request from the user comprises a first step of determining whether the requested stream data exists in its database; A second step of transferring the service to the stream pool having the requested stream data if the requested stream data does not exist in its database as a result of the determination of the first step; And And a third step of allocating a serviceable stream processor by applying a dynamic load balancing module if the requested stream data exists in its database as a result of the determination of the first step. The method of claim 3, wherein The first step is, When a user requests a service by transmitting his specification and information about a stream to be serviced to a stream pool to which he is subscribed, the stream pool receiving the service determines whether the requested stream data exists in his stream database. On-demand information providing service method characterized in that. The method according to claim 3 or 4, The third step, A fourth step of checking whether a serviceable stream processor exists if the requested stream data exists in its database; A fifth step of providing a service to the user by allocating the service to the serviceable stream processor if a serviceable stream processor exists as a result of the checking of the fourth step; And And a sixth step of providing a service to a user by allocating a service to the serviceable stream processor after moving the load by applying dynamic load balancing if the serviceable stream processor does not exist as a result of the determination in the fourth step. On-demand information providing service method characterized in that. The method according to claim 3 or 4, The third step, A fourth step of checking whether a serviceable stream processor exists if the requested stream data exists in its database; A fifth step of providing a service to the user by allocating the service to the serviceable stream processor if a serviceable stream processor exists as a result of the checking of the fourth step; A sixth step of determining whether a serviceable stream processor exists after the load is moved by applying a dynamic load balancing technique in the stream pool if the serviceable stream processor does not exist as a result of the determination of the fourth step; A seventh step of providing a service to the user by allocating a service to the serviceable stream processor when the serviceable stream processor exists as a result of the determination of the sixth step; And As a result of the determination of the sixth step, if there is no serviceable stream processor, an eighth step of providing a service to a user by allocating a service to the serviceable stream processor after shifting the load by applying a dynamic load balancing technique among the stream pools; On-demand information providing service method comprising a. The method according to claim 3 or 4, The third step, Upon receiving the stream request, the dynamic load balancing module checks whether there is a serviceable stream processor among its stream processors that meets the quality of service conditions of the stream while having IDLE or LIGHT LOADED. 4 steps; A fifth step of performing a service by allocating a service for the stream to the serviceable stream processor when the serviceable stream processor exists as a result of the checking of the fourth step; A sixth step of re-confirming whether there is a serviceable stream processor after moving the load in the stream pool by obtaining normalized values for load movement if there is no serviceable stream processor as a result of the checking in the fourth step; A seventh step of starting a service by allocating a service for the stream to the serviceable stream processor if a serviceable stream processor exists as a result of the checking of the sixth step; And As a result of the checking in the sixth step, if there is no serviceable stream processor, the normalized values for load movement are obtained, the load is moved between the video stream pools, and then the service is allocated to the stream processor to start the service. On-demand information providing service method comprising the eighth step.

Images