KR20110063285A - Testbed system for dynamic media service composition and experiment method for media service composition - Google Patents

Abstract

PURPOSE: A test bed system for a dynamic media service synthesis and a media service synthesis experiment method is provided to integrally operate dispersed computing/networking services by using resource of the test bed. CONSTITUTION: A test bed management server(300) manages a media resource group(100) and a network resource group(200). The test bed management server grasps an operation state of the network resource group and the media resource group. The test bed management server assigns resources for a test. A test bed experiment control server(400) controls a service composition. The test bed experiment control server corresponds to detected services and resources.

Description

Testbed system for dynamic media service composition and experiment method for media service composition}

The present invention relates to media service synthesis in an Internet environment, and more particularly, to a testbed system and a method for experimenting on media service synthesis for dynamic media service synthesis.

In recent decades, the amount of media content delivered over the Internet has increased dramatically. As the personal demand for realistic media services increases, the need for flexible multimedia systems increases. A flexible multimedia system is a system that can understand media content and frameworks scattered across various computing / networking environments, synthesize media content, and deliver media content continuously. In building flexible multimedia applications, media service synthesis research has been conducted to accommodate the fundamental memorial of Service-Oriented Computing (SOC). This paradigm has greatly influenced the expansion of monotonous multimedia applications to more flexible component-based applications. Media service composition automatically configures multimedia applications at runtime by adjusting the composable services according to a given service graph. For reference, media service composition inherits the basic principles and methodology from web service composition. Media service synthesis is similar to web service synthesis in that it consists of a number of composable services, but media service synthesis is handled more carefully than web service synthesis in that it must ensure continuity of media data delivery under time and resource constraints. You must lose.

Media service synthesis research requires a testbed that can "configure computing / networking resources the way an experimenter wants" to test and verify this in a realistic way. For reference, simulation virtually supports media service synthesis experiments based on quantitative mathematical / logical models, so it is difficult to accurately reflect the various variables derived from the real Internet environment. Representative studies for testbeds include PlanetLab, VINI and OMF. PlanetLab provides a global testbed for experimenting with new protocols and applications using computing resources distributed across the Internet in OSI L3 and above. Virtual Network Infrastructure (VINI) supports networking virtualization at the OSI L2 layer, along with the features provided by PlanetLab, enabling more extensive networking protocol experiments. The Orbit Management Framework (OMF) provides an environment where ideas can be validated in a networking testbed consisting essentially of a wireless LAN. These testbeds provide the deployment of experimental tools, resource allocation and monitoring, and measurement of experimental results. However, there is still a need for a function of dividing internally complex and tightly coupled media processing and transmission functions by element, and flexibly and balancingly connecting computing / networking services according to resource status.

An object of the present invention is to provide a test bed system for effectively supporting an experiment for dynamically synthesizing distributed media-oriented services.

In order to solve the above technical problem, a test bed system for synthesizing a dynamic media service includes a media resource set, which is a set of computing resources for performing media data processing; A network resource set, which is a set of computer resources that provide programmable networking and network virtualization; A server for managing the media resource set and the network resource set, the server comprising: a test bed management server configured to determine an operation state of the media resource set and the network resource set and to allocate resources for a test; And a server for controlling a service synthesis experiment. A testbed experiment control server that interprets a service dependency graph to match found services and resources, activates a service, and connects each service instance to generate a service instance dependency graph. Characterized in that it comprises a.

Here, the computing resources for performing the media data processing may include a camera, a display device, a mass storage, a GPU-based computing device.

In addition, the testbed system may further include an experimenter interface that receives an experimenter to generate the service dependency graph.

In addition, the media resource set may include media nodes providing media resources and a media RA manager managing the media nodes.

The media resource set may include a sliver that virtualizes the computing resource.

In addition, the network resource set may be a set of OpenFlow-based PC resources that provide programmable networking and network virtualization.

The test bed management server may further include a slice manager that allocates, adjusts, and retrieves slices for the media resource set and the network resource set, registers information of experimenters using slices, and registers slice setting information. can do.

The test bed management server may include a test bed resource manager which maintains a resource list and a current state of the media resource set, the network resource set.

The test bed experiment control server may include a service monitoring / tuning module that monitors whether service instances operate normally and notifies an experimenter of an error that has occurred.

In order to solve the above technical problem, the media service synthesis experiment method in the test bed system, generating a service dependency graph by connecting the services according to the input of the experimenter; Allocating resources of the media resource set and the network resource set required for the services; Associating the allocated resources with the services according to the service dependency graph; And executing the services and connecting each service instance to generate a service instance dependency graph.

In order to solve the above technical problem, there is provided a computer-readable recording medium recording a program for executing the above-described media service synthesis experiment method.

According to the present invention, by providing a system and method for integrating and operating distributed computing / networking services using the resources (processing capacity, storage space, network bandwidth, etc.) provided by the test bed, the experimenter The synthesis experiment can be conveniently performed.

1 is a diagram illustrating a structure of a component service.
2 is a diagram illustrating a service dependency graph and a service instance dependency graph.
3 illustrates a media service synthesis example using a testbed system according to an embodiment of the present invention.
4 illustrates a configuration of a testbed system for dynamic media service synthesis according to an embodiment of the present invention.
5 shows an example of media service synthesis.
6 is a flowchart illustrating a procedure of synthesizing a media service using the above-described testbed system according to an embodiment of the present invention.
7 illustrates an example of an interface screen provided by the experimenter interface 500.
8 shows a resource allocation procedure.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the following description and the accompanying drawings, substantially the same components are denoted by the same reference numerals, and redundant description will be omitted. In addition, in the following description of the present invention, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

First, a service model and a media service synthesis scenario for media service synthesis to which the present invention is applied will be described.

In the present specification, a 'service' means a component service, and a component service is used to access a specific function provided by software or hardware through a standardized interface to utilize the function according to a given access / control right. Recyclable unit module. 1 is a diagram illustrating a structure of a component service. In general, a component service is described with three attributes as shown in FIG. 1, which includes 1) a name representing a main function of a component service, and 2) a handler for manipulating the main function of the service by adjusting predefined control parameters. 3) it consists of interfaces for exchanging processed data between services through standardized interaction protocol. Since the quality of service is highly dependent on the amount of resources to be supplied, stable service operation requires a clear declaration of the minimum resources required for the service to perform its functions.

Component services are classified into computing-centric services and networking-centric services, depending on their purpose. Computing-centric services are responsible for acquiring, processing, transmitting and presenting media using CPU / GPU capabilities, storage, cameras and displays. Typical service examples include media producer service, media consumer service, transcoding service, video composition service, and video tiling service. Networking-centric services must take advantage of the virtualized networking capabilities provided by testbeds to ensure continuous and reliable delivery of processed data between computing-centric services. For example, a forwarding service that forwards data from one node to another node, a replication service that distributes copies of the data to multiple destinations, and a NAT / firewall traversal service to ensure data delivery by bypassing firewalls or NATs. -centric services.

Service synthesis allows the experimenter to perform complex tasks consisting of unit tasks. Service composition is described by the information of component services and their execution dependencies. In the present invention, the execution dependency is abstractly expressed in the form of a service dependency graph. The service dependency graph is represented by a set of service pairs connected by dividing lines. Here, the frankincense segment indicates the calling order of the services, and the starting point and the arrival point of the frankincense segment indicate the preceding service and the following service, respectively. According to the service dependency graph, it runs the specified component services on the computing / networking resources allocated from the testbed system according to the present invention and interconnects them. 2 is a diagram illustrating a service dependency graph and a service instance dependency graph. As shown in FIG. 2, the driven component service is called a service instance, and the set of connected service instances according to the service dependency graph is called a service instance dependency graph.

3 illustrates a media service synthesis example using a testbed system according to an embodiment of the present invention. Users present media obtained from on-demand content services and live capturing services using mobile display services, digital TV services, and large networked display services. In the case of mobile displays, there is a limited range of computing / networking capabilities that are permissible, making it difficult to receive these media directly. For digital TVs, content can generally be received directly within the acceptable range. Have the ability to In the case of a large network display, a large amount of contents can be directly received, but the contents must be divided and transmitted according to the size of the fragmented displays constituting the display. In the case of mobile display and digital TV, a media composition of on-demand content service and live capturing service is synthesized into a single stream using a video composition service, and transcoding service is limited to a stream delivered to mobile display. Media media quality is adjusted down. The path of the media stream to the large network display is slightly different. Streams obtained from the on-demand content and live capturing services go through a video composition service and a caching service that replicates and archives the data, and then divides the video tiling service into the size of the individual displays that make up the network display. Pass to display service.

4 illustrates a configuration of a testbed system for dynamic media service synthesis according to an embodiment of the present invention.

In the test bed system according to the present embodiment, a media resource aggregate 100, a network resource aggregate 200, a testbed management server 300, a testbed experiment The control server (Testbed Experiment Control Server) 400, including the experimenter interface 500 is made.

The media resource set 100 is a set of computing resources that perform media data processing, and includes a camera, a display device, a mass storage, and a GPU-based computing device.

Network resource set 200 is a collection of computer resources that provide programmable networking and network virtualization.

The test bed management server 300 is a server managing the media resource set 100 and the network resource set 200, and grasps the resource list and the operation state of the media resource set 100 and the network resource set 200. Adjust test bed usage rights for the experimenter, detect false experiments, and allocate, adjust and reclaim resources for testing.

The testbed experiment control server 400 is a server for controlling a service synthesis experiment, and interprets a service dependency graph to match found services and resources, activates a service, and connects each service instance to a service instance dependency graph. Create

The experimenter interface 500 is an experimenter interface that allows an experimenter to receive an input to generate a service dependency graph for a service synthesis experiment and to select necessary resources.

Hereinafter, each configuration of the test bed system will be described in more detail.

The media resource set 100 is composed of computing resources suitable for real-time and mass media processing, such as a camera, a display device, a mass storage, and a GPU-based computing device. As shown by # 1, # 2, and # 3, a plurality of media resource sets 100 may exist.

The media resource set 100 is composed of media nodes 120 that provide media resources, and a media RA manager 110 that manages the media nodes 120.

Media node 120 may be a personal computer (PC) with computing resources. Hardware substrates provide computing resources such as processors, storage and graphics accelerators. These computing resources can be virtualized to effectively support multiple, simultaneous experiments. Media node 120 may be divided into three types depending on the level of virtualization. First, non-virtualized media nodes are ideal for performing I / O tasks such as GPU-based graphics acceleration, video capture cards, and displays. Second, virtualized media nodes are primarily suited for tasks that require computing and data storage capabilities. Third, cloud media nodes can be used for tasks that require large scale virtual computing resources composed of distributed computing resources.

The node manager 121 virtualizes a computing resource to be used by the service to make it a sliver 122, and executes a computing service requested by the test bed experimental control server 400 thereon. Here, the sliver means a single virtual machine created through node virtualization on each node in PlanetLab. In addition, the computing service is a component service, and refers to a service using a virtual computing node (a virtualized computing resource in one node) for processing of media, for example, a transcoding service, a caching service, and a compression. ) Services, etc.

The resource registries of internal elements of the media RA manager 110 store information of nodes constituting the media resource set 100. When the request of the testbed management server 300 comes, the media RA manager 110 allocates specific nodes as resources for the experiment. In order to observe whether the nodes are operating stably, the RA monitor checks the performance change of the nodes from time to time and reports the result to the TB resource monitor of the test bed management server 300. .

Network resource set 100 is a set of OpenFlow-based PC resources that provide programmable networking and network virtualization. For OpenFlow, see N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson, J. Rexford, S. Shenker, and J. Turner, "OpenFlow: Enabling innovation in campus networks," ACM SIGCOMM Computer Communications Review, vol. 38, no. 2, pp. 69-74, Mar. 2008.]. Similar to the media resource set 100, the network resource set 100 is composed of network nodes 220 that provide network resources and a network RA manager 210 that manages the network nodes 220. The hardware substrate provides a variety of networking resources, such as hardware accelerated wired network interfaces and wireless network interfaces using NetFPGA. For NetFPGA, see G. Gibb, J.W. Lockwood, J. Naous, P. Hartke, and N. McKeown, "NetFPGA-An open platform for teacing how to build Gigabit-rate network switches and routers," IEEE Trans. on Education, vol. 51, no. 3, pp. 364-369, Aug. 2008.]. These networking resources can also be virtualized to effectively support multiple, simultaneous experiments. Programmable networking allows you to control the data flow of a given network according to the intention of the experimenter.

The node manager 221 virtualizes the networking resource to be used by the service to make it a sliver 222, and executes the networking service requested by the test bed experimental control server 400 thereon. Here, the networking service refers to a service that uses a virtual networking link (a virtualized networking resource in one node) to connect a virtual computing node as a component service. For example, a forwarding service, a multicast ( This includes multicast services and congestion control services.

The resource registries of the internal elements of the network RA manager 210 store information of nodes constituting the network resource set 200. When a request of the testbed management server 300 comes, the network RA manager 210 allocates specific nodes as resources for experimentation. In order to observe whether the nodes are operating stably, the RA monitor checks the performance change of the nodes from time to time and reports the result to the TB resource monitor of the test bed management server 300. .

The test bed management server 300 may include a slice manager 310, a test bed resource manager 320, a test bed resource monitor 330, and a test bed management user interface ( TB Mgmt UI.) 340.

The slice manager 310 allocates, adjusts, and retrieves slices of all the media resource sets 100 and the network resource sets 200. Here, the slice refers to a network of computing resources and networking resources capable of executing one experiment. The slice manager 310 registers the information of the experimenters who use the slice in the user registry and registers the slice setting information in the slice registry. The test bed resource manager 320 maintains a resource list and a current state of the media resource set 100 and the network resource set 200 in the TB Resource Registry. The test bed resource monitor 330 monitors the status of the resources provided by the media resource set 100 and the network resource set 200. The test bed management user interface 340 is an interface that allows administrators to check and control the status of resources.

The test bed experiment control server 400 abstracts the interfaces of the resource sets so that an experimenter can control the service for the experiment independently of the characteristics of the different resource sets, thereby providing a high level experiment technique. .

Basically, an experiment consists of one or more slice (s), and one slice is used to run one synthesis service. In this case, the synthesis service may be composed of a combination of component services that operate across various resource sets. In order to create a synthesis service, an experimenter is assigned a slice, and then uses the slice to instantiate the component services (distribute the component services to a specific resource, configure them according to given parameters, and execute them), and synthesize them. (A process of interconnecting component services to transfer data according to a given workflow), and then constructing the resulting composite service to perform part or all of the planned experiment.

The test bed experiment control server 400 includes an analysis module 440 for service synthesis, a service discovery / matching module 410, a service deployment / connection module 420, and a service monitoring / tuning module 430. The testbed experiment control server 400 interacts with the experimenter through the experimenter interface 500. The interpretation module 440 interprets the service dependency graph received from the experimenter interface 500, and the service discovery / matching module 410 discovers the necessary service, and thus, the found services, the media resource set 100, and the network resource. Match the resources in the set 200 as appropriate. The service placement / connection module 420 activates services corresponding to the selected resource, and actually connects each service instance to generate a service instance dependency graph. Information about the resources used for the experiment is stored in the Experiment Registry. The service monitoring / tuning module 430 monitors whether service instances are operating normally, and immediately notifies the experimenter of the generated error through the experimenter interface 500. Alternatively, the experimenter uses the service monitoring / tuning module 430 to finely adjust the control parameters of the services according to the defined rule.

5 shows an example of media service synthesis. The test bed's resource set (RA) consists of nodes for media servers, adapters, and displays. Nodes for media server perform uncompressed HD capturing service, DXT (DirectX Texture) compression service, MPEG compression service from HD cameras, nodes for adapters perform MPEG2 decoding service, nodes for display are network display Perform the service.

FIG. 6 is a flowchart illustrating a procedure of synthesizing a media service using the above-described testbed system, and FIG. 7 illustrates an example of an interface screen provided by the experimenter interface 500. Referring to FIG.

In operation 610, the experimenter interface 500 displays a list of services supported by the test bed system (710 of FIG. 7), selects a set of services according to the experimenter's input, and then connects the services by using a frankincense segment. A graph (720 of FIG. 7) is generated.

In operation 620, the experimenter interface 500 displays a list of available resources (730 of FIG. 7), and the experimenter confirms this. In order to request a resource required for the experiment, the experimenter may select a desired node through the experimenter interface 500 and request the testbed experiment control server 400 for the required amount of resources. At this time, the amount of resources may be quantitatively indicated by numerical values of computing capability, storage capability, and networking capability.

In operation 630, the test bed experiment control server 400 requests the test bed management server 300 to allocate the resources of the media resource set 100 and the network resource set 200 to allocate the requested resources. . 8 shows a resource allocation procedure. Referring to FIG. 8, the test bed management server 300 communicates with the RA managers 110 and 210 and performs 'Resource Discovery' inquiring whether it can provide necessary resources. The test bed management server 300 checks whether the experimenter has permission to access the requested resource (Check Access Authorities). If authorized, the RA managers 110 and 210 create a set of slivers and deliver it to the testbed experimental control server 400 (Allocate a set of resources). ).

In operation 640, the test bed experiment control server 400 searches for a service repository including a software package of services selected by an experimenter.

In operation 650, the testbed experiment control server 400 matches the services with the resources prepared according to the service dependency graph.

In operation 660, the test bed management server 300 installs a software package of services on the prepared resource. The test bed experiment control server 400 executes the installed services. The executed service corresponds to the service instance.

In step 670, the testbed experiment control server 400 informs the service instances of their neighboring service instances according to the execution order of the services. Service instances receive their addresses and then communicate with each other to exchange media data so that services are connected (creating a service instance dependency graph). 740 of FIG. 7 is a generated service instance dependency graph. For example, in LiveCapturingService @ node-camera-h and DecoderSevice @ node-decoder1, LiveCapturingService and DecoderSevice are services, and node-camera-h and node-decoder1 following @ are slivers in which the service is running. Means 'connected'.

In operation 680, the test bed experiment control server 400 monitors whether the synthesized services play a given role. The obtained monitoring information may be referred to other services to perform 'service tuning'.

In step 690, the services are reset using the monitoring result to continue the service synthesis according to the change of the test bed system (resources, services).

Meanwhile, the above-described embodiments of the present invention can be written as a program that can be executed in a computer, and can be implemented in a general-purpose digital computer that operates the program using a computer-readable recording medium. The computer-readable recording medium may be a magnetic storage medium (for example, a ROM, a floppy disk, a hard disk, etc.), an optical reading medium (for example, a CD-ROM, DVD, etc.) and a carrier wave (for example, the Internet). Storage medium).

So far I looked at the center of the preferred embodiment for the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

Claims (11)

Testbed system for dynamic media service synthesis
A media resource set, which is a set of computing resources that perform media data processing;
A network resource set, which is a set of computer resources that provide programmable networking and network virtualization;
A server for managing the media resource set and the network resource set, the server comprising: a test bed management server configured to determine an operation state of the media resource set and the network resource set and to allocate resources for a test; And
As a server for controlling service synthesis experiments, a testbed experiment control server that interprets a service dependency graph to match found services and resources, activates a service, and connects each service instance to generate a service instance dependency graph. Test bed system comprising a. The method of claim 1,
The computing resources for performing the media data processing includes a camera, a display device, a mass storage, a GPU-based computing device. The method of claim 1,
And an experimenter interface receiving an experimenter to generate the service dependency graph. The method of claim 1,
The media resource set includes a media node for providing a media resource and a media RA manager for managing the media nodes. The method of claim 1,
And the media resource set includes a sliver that virtualizes the computing resource. The method of claim 1,
And said network resource set is a set of OpenFlow-based PC resources that provide programmable networking and network virtualization. The method of claim 1,
The test bed management server may include a slice manager that allocates, adjusts, and retrieves slices for the media resource set and the network resource set, registers information of experimenters using slices, and registers slice setting information. Characterized test bed system. The method of claim 1,
The test bed management server includes a test bed resource manager which maintains a resource list and a current state of the media resource set and the network resource set. The method of claim 1,
The testbed experiment control server includes a service monitoring / tuning module for monitoring whether service instances are operating normally and notifying an experimenter of an error that has occurred. A media service synthesis experiment method in the test bed system according to claim 1,
Connecting services according to an experimenter's input to generate a service dependency graph;
Allocating resources of the media resource set and the network resource set required for the services;
Associating the allocated resources with the services according to the service dependency graph; And
Running the services and connecting each service instance to generate a service instance dependency graph. A computer-readable recording medium having recorded thereon a program for executing the media service synthesis experiment method according to claim 10.

Images