KR101040171B1 - Scalable Grid Resource Management based on WBEM with UPnP resource discovery - Google Patents

Abstract

The present invention relates to a WBEM-based scalable grid resource management apparatus and method including UPnP resource discovery, and includes UPnP resource discovery as a structure for distributed resource management based on UPnP and WBEM (Web-based Enterprise Management). Perform scalable grid resource management based on WBEM.

According to the present invention, a hierarchical resource management model for the grid structure resulting from the grid forum, the important elements of the presented structure are divided into passive and active elements. And the function of this element is also the same. The main difference in the proposed structure is that WBEM is used as a management framework, resource discovery is based on WBEM / UPnP, and SLA providers were developed during the negotiation phase. In addition, CIM further extends to managed objects.

Grid, Grid Resource Management, WBEM, UPnP, Resource Discovery

Description

Scalable Grid Resource Management based on WBEM with UPnP resource discovery}

The present invention relates to a structure for distributed resource management based on Universal Plug and Play (UPnP) and Web-based Enterprise Management (WBEM), and more specifically, a scalable WBEM framework and extended UPnP-based resource discovery. The present invention relates to a WBEM-based scalable grid resource management apparatus including UPnP resource discovery.

An important aspect of resource management is the definition of a generic framework for negotiation between different vendors with different policies. Universal Plug and Play (UPnP) was established as a technology that controls and transfers data between network resources and enables near-network networking without conflict.

WBEM can provide the capability for a well-integrated industry with standards-based tools, and UPnP is currently used to use network equipment in a plug-and-play (PnP) method without the need for manual configuration and installation. Is the definition and standard.

1) Resource Management in Grid Computing

An important part of the grid organization is the ability to share resources for quality assurance services. In a grid environment, there are already many resources that include providing services to users. Each owner has their own personal policy requirements. Resource management, therefore, is becoming increasingly complex, and many organizations have made consistent policies across policy domains. In a large grid network, the most important requirements in resource management are to discover, negotiate, and allocate the resources that users require.

[Grid]

Computer grids are hardware and software based that provide reliable, consistent, widespread, and inexpensive access to the highest level of computing power. The grid equally integrates users and resources in different control domains. It also focuses on basic issues such as resource access, authentication, and resource discovery, and is based on protocols and interfaces for various purposes. These protocols and interfaces are important because they are standardized and published.
[Grid Computing]

delete

Grid computing is being considered as the next step in distributed computing. Based on widespread Internet standards, grid computing can join organizations in a more efficient way to share computing and information resources and in terms of security.

2) WBEM CIM

WBEM is a set of standardized techniques including (i) CIM 101, (ii) CIM-XML 106, and (iii) Hyper-Text Transport Protocol (HTTP) 105 approach. CIM-XML is an XML-concatenated xml that uses payload and HTTP as the transport. CIM-XML is a CIM message that defines every interaction between CIM products. The purpose of xmlCIM is to make XML grammar by DTD (Document Type Definition). It can also be used to represent CIM declarations and CIM messages for use by the CIM protocol. The WBEM client 107 may request the WBEM server 108 using the CIM schema 101. The appropriate Appropriate Provider 109 is then initiated from the WBEM server, and management can obtain information from the resource 110 upon request.

[WBEM]

The Distributed Management Task Force (DMTF) WBEM is a management standard for distributed computing environments.

WBEM is a technology that integrates well with standards-based technologies and can provide good performance for data delivery. It also facilitates data exchange between opposing technologies and platforms.
[CIM]

delete

The DMTF Common Information Model (CIM) 101 structure aims at the management of networks and systems using the conceptualization and basic structure of the object-oriented paradigm. CIM is composed of individual elements and in a way that manages the environment as a collection of related systems. The CIM includes a specification 103 and a schema 104. The CIM specification 103 defines the integration between different management models, and the CIM schema 104 provides a practical model description.

3) UPnP

UPnP architecture relies on peer-to-peer networking between PCs, networked devices, or wireless devices. The UPnP architecture is a distributed, open architecture based on TCP / IP, UDP and HTTP. A collection of standards and definitions for using networked devices in a plug-and-play manner without manual configuration or setup. Designed for the purpose of automatic discovery for the operation of device categories from a wide range of vendors, from invisible networks, and without special configuration. The UPnP architecture allows devices 202 to automatically access the network, Obtain an address, pass device characteristics, and learn about other device characteristics that are current. Finally, the device can seamlessly disconnect from the network without leaving any state when it leaves the network smoothly. Device discovery, IP address acquisition, and other control messages are operated by the control point 201. Some devices may include a sub-device 204 or a service 203. UPnP is a technology that allows users to seamlessly connect to a network to transfer and control data between networked devices.

[UPnP]

UPnP is a set of computer network protocols advertised by the UPnP Forum. The purpose of UPnP is to provide a seamless and seamless connection to the networks used at work and at home. UPnP defines and publishes the UPnP device control protocol to achieve an Internet-based communication standard. The term UPnP is derived from Plug-and-paly and is a technology for connecting devices directly to computers.

[UPnP Protocol]

The first step in UPnP is discovery. When a new device is added to the network, the UPnP discovery protocol relies on the device to advertise for service to control points in the network. When a control point is newly added to the network, the UPnP discovery protocol follows the control point to find the device of interest on the network.

Then, the technical stage is made. Even after the control point finds the device, it knows little about the device. In order for the control point to obtain more information, the control point must return the device's description from the URL provided by the device in the discovery message. UPnP technology for devices is represented in XML and includes vendor information and manufacturer information.

The control is executed after the above description. After the control point returns a description of the device, the control point can send an action to the device's service. To do this, the control point must send an appropriate control message to the control URL for the service (which provides the device description). Control messages can also be represented in XML.

The next step is UPnP networking for event notification or “eventing”. The service may initiate an update when various variables change or when the control point agrees to receive information. The service initiates the update by sending an event message.

The final step is that if the device has a URL for presentation, the control point loads the page according to the characteristics of the web browser and page, and the user to control the device and the current state of the device, and retrieves the page from this URL. Can be returned.

In the prior art, one of the criteria for the grid structure is whether resources can be shared with quality assurance. In a grid environment, there are many resources involved in the services provided to the user. And there is a problem that the resource manager must be able to manage huge amounts of data. Another problem with resource management in grid networks is the difference between resources. According to the greatness of the grid network, the most important requirement of the resource management system is to discover, negotiate, and allocate the use of the necessary resources.
Accordingly, the present invention has been made to solve the above-described problems, and an object of the present invention is to provide a common model and protocol for controlling and monitoring resources from various sources defined by the WTFM platform and resource-independent DMTF standard. Incorporating WBEM technology into grid computing to overcome the shortcomings seen in previous technologies, WnM-based, including UPnP resource discovery, allows UPnP to be used for resource discovery and negotiations to be run through service level agreements (SLAs). The purpose is to provide a scalable grid resource management apparatus.

delete

In a grid computing environment, a grid resource 317 necessary for managing grid resources (services provided by a device or a device) 317 distributed based on Universal Plug and Play (UPnP) and Web-based Enterprise Management (WBEM) is provided. WBEM-based grid resource management device that includes resource discovery by UPnP for discovering, negotiating, and allocating a grid resource 317 that can be allocated and used for user tasks and jobs in a grid computing environment. ); A UPnP provider 311 that provides the discovery and detailed information acquisition and control of the grid resource 317; A resource provider 312 performing a resource allocation and provision function so that the grid resource 317 found by the uPnP provider 311 may be efficiently used in a grid computing task and a job; A scheduler 302 that calculates one or multiple schedules to determine the execution order of each job for the resource 317; A domain control agent accommodating a resource to be executed among the grid resources 317; A deployment agent that implements a schedule through negotiation with the domain control agents, obtains a grid resource 317, starts a task, and performs mapping to an SLA provider 305; An admission control agent that determines whether to accept to complete each job requested by the scheduler 302; A CIM client delivering the job to a resource management system for execution of a user; A monitor for tracking the progress of the job according to the determined execution order by the scheduler 302; It is characterized in that it comprises a; job control agent to watch the job through the resource management system, or to act as a proxy for the user and a continuous control point for the job (job).

According to the present invention as described above, the hierarchical resource management model for the grid structure that is the result of the grid forum, the important elements of the proposed structure is divided into passive and active elements. And the function of this element is also the same. The main difference in the proposed structure is that WBEM is used as a management framework, resource discovery is based on WBEM / UPnP, and SLA providers were developed during the negotiation phase. In addition, CIM further extends to managed objects.
As described above, the WBEM and UPnP based distributed resource management apparatus according to the present invention can be successfully applied in grid computing and grid structure, and the scalability, feasibility and reliability of integration can be provided by the WBEM and UPnP architectures. Negotiation between the client and the resource management application can be handled through the SLA provider of WBEM.

Hereinafter, the WBEM and UPnP-based distributed grid resource management structure according to the present invention will be described in more detail with reference to the accompanying drawings.

The present invention integrates WBEM technology into grid computing to overcome the drawbacks of previous technologies from common models and protocols for controlling and monitoring resources from various sources defined by the WTFM platform and resource-independent DMTF standards. UPnP is used for resource discovery, and negotiations are run through SLAs.

[SLA]

The SLA is a formal negotiated contract between two parts. A contract exists between a consumer and a service provider or service provider. It records a general understanding of services, priorities, responsibilities, guarantees, etc. with the main purpose of agreeing to the level of service. For example, it may specify the level of availability, available standards, performance, operation or other attributes of services such as billing and penalties in case of violation of the SLA.

The proposed grid resource management architecture includes two types of components: passive components and active components.

First, the passive component is:

Resources (services provided by the device or any device) 317 are used according to the time owner. And it may or may not start again.
Tasks are both destructive (consumer) resources, and include both traditional destructive (consumable) and nondestructive tasks.

delete

 A job consists of hierarchical entities and sub jobs or tasks.

 Schedules map tasks to resources that are out of date.

The active component structure is:

The scheduler 302 calculates one or several schedules for the list of jobs to be entered, where the unit of work is the job. That is, one or more schedules are calculated to determine the execution order of each job for each grid resource 317.

The information service runs as a database for the described item. It is also easily managed by the CIM Repository 309.

A domain control agent may accommodate a resource to be executed among the grid resources 317.

Deployment Agent implements the schedule through negotiation with domain control agents. The deployment agent acquires resources and starts the task. And included in the SLA provider 305 obligation.

-The Admission Control Agent can determine whether the system can supply additional jobs to the system and reject it or defer depending on when it will be applied to the system. This function is also included in the SLA provider 305 obligation. That is, it determines whether to accept each job requested by the scheduler 302 to complete.

-The user is a CIM client 310 which entrusts the job to the resource management system for execution.

The monitor tracks the progress of the job according to the determined execution order by the scheduler 302. This functionality is similar to that of Performance and Fault manager 307 and 303. The execution and error managers 307 and 303 are UPnP provider 311, CLI provider 313 and SNMP provider 314, and CIMOM 304 or scheduler 302 according to the remapping effect of the job. Get the status of a job according to the return value of.

Job Control Agent has the role of watching job through the system, acting as proxy for user, and acting as a continuous control point for job.

Integration between components of distributed resource management is similar to integration in a hierarchical model. The client requests a job from the CIMOM 304 called the SLA provider. The SLA provider examines the resource requirements for the job and determines whether the job is safely accepted by the system. If possible, the SLA provider invokes a scheduler 302 that can obtain information about the resources available through the resource provider 312. The resource provider obtains information on these available resources through the UPnP provider 311, the CLI provider 313, and the SNMP provider 314, and makes a decision about the current state and the usefulness of the resource. Existing information in the CIM repository 308 is updated according to the findings and determined results.
The scheduler 302 then calculates the set of mappings and then maps to the SLA provider. The SLA provider negotiates with the resource provider 313 the resources indicated by the schedule. Then prepare a reservation for the resource. This reservation is delivered through the CIMOM 304 and is in accordance with the schedule determined by the scheduler 302. CIMOM 304, which is equivalent to resource provider 312, initiates the task. The performance manager 307 and the fault manager 303 keep track of the progress of the job. And, depending on the measured performance result, it may be reorganized or the schedule may be reconfigured.

WBEM-based scalable grid resource management apparatus and method including UPnP resource discovery according to the present invention described above is not limited to the above-described embodiment, without departing from the gist of the invention claimed in the claims below Anyone having ordinary knowledge in the field of the present invention has the technical spirit to the extent that it can be variously modified and implemented.

1 shows a typical WBEM structure.

2 illustrates a general UPnP structure.

3 is a diagram illustrating a WBEM and UPnP based distributed grid resource management structure according to the present invention;

Claims (7)

Discover, negotiate, and manage the grid resources 317 necessary for management of multiple grid resources 317 distributed based on Universal Plug and Play (UPnP) and Web-based Enterprise Management (WBEM) in a grid computing environment, In the WBEM-based grid resource management apparatus including resource discovery by UPnP to allocate, A grid resource 317 that can be allocated and used for user tasks and jobs in the grid computing environment; A UPnP provider 311 that provides the discovery and detailed information acquisition and control of the grid resource 317; A resource provider 312 performing a resource allocation and provision function so that the grid resource 317 found by the uPnP provider 311 can be efficiently used in a grid computing task and a job; A scheduler (302) for calculating one or more schedules to determine an execution order of each job for each grid resource (317); A domain control agent accommodating a resource to be executed among the grid resources 317; A deployment agent that implements a schedule through negotiation with the domain control agents, obtains a grid resource 317, starts a task, and performs mapping to an SLA provider 305; An admission control agent that determines whether to accept the job to complete each job requested by the scheduler 302; A CIM client delivering the job to a resource management system for execution of a user; A monitor for tracking the progress of the job according to the determined execution order by the scheduler 302; And A job control agent that monitors a job through the resource management system or serves as a proxy for the user and a continuous control point for the job. UPnP WBEM-based scalable grid resource management device with resource discovery. delete The method of claim 1, The domain control agent controls resources through a UPnP provider 311, a command line interface (CLI) provider 313, a simple network management protocol (SNMP) provider 314, and a resource provider 312. WBEM-based scalable grid resource management device with resource discovery. The method of claim 1, The monitor uses the UPnP provider 311, the CLI provider 313, the SNMP provider 314, and the resource provider 313, and uses the CIMOM (CIM Object Model) 304 according to the remapping effect of the job. WBEM-based scalable grid resource management apparatus including UPnP resource discovery, which tracks the progress of the job by acquiring the status of the job according to the return value of the scheduler 302 . The method according to claim 1 or 4, The SLA provider 305 examines the resource's request for a job, determines whether the job is safely accepted by the system, and if so, through the resource provider 312 Invoke the scheduler 302 to obtain information about available resources, The resource provider 312 obtains information on the available grid resource 317 through the UPnP provider 311, the CLI provider 313, and the SNMP provider 314, and the current state of the grid resource 317. And WBEM-based scalable grid resource management device, including UPnP resource discovery, characterized in that the decision making and usability. The method of claim 5, The SLA provider 305 negotiates with the resource provider 312 a grid resource 317 indicated by a schedule, and prepares a reservation for the grid resource 317. UPnP resource discovery WBEM-based scalable grid resource management device, including. The method of claim 6, The reservation for the grid resource (317) is delivered through the CIMOM (304), WBEM-based scalable grid resource management apparatus including UPnP resource discovery, characterized in that according to the schedule of the scheduler (302).

Images