WO2015024368A1 - 一种面向构件的混合型云操作***体系结构及其通信方法 - Google Patents

一种面向构件的混合型云操作***体系结构及其通信方法 Download PDF

Info

Publication number
WO2015024368A1
WO2015024368A1 PCT/CN2014/071233 CN2014071233W WO2015024368A1 WO 2015024368 A1 WO2015024368 A1 WO 2015024368A1 CN 2014071233 W CN2014071233 W CN 2014071233W WO 2015024368 A1 WO2015024368 A1 WO 2015024368A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
component
queue
write
read
Prior art date
Application number
PCT/CN2014/071233
Other languages
English (en)
French (fr)
Inventor
王恩东
张东
刘正伟
亓开元
刘俊朋
郭峰
刘成平
高飞
朱波
Original Assignee
浪潮电子信息产业股份有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 浪潮电子信息产业股份有限公司 filed Critical 浪潮电子信息产业股份有限公司
Priority to JP2015543294A priority Critical patent/JP2015537307A/ja
Priority to EP14833104.4A priority patent/EP2899947A4/en
Priority to US14/534,735 priority patent/US9307017B2/en
Publication of WO2015024368A1 publication Critical patent/WO2015024368A1/zh

Links

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • H04L67/1001Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1004Server selection for load balancing
    • H04L67/1008Server selection for load balancing based on parameters of servers, e.g. available memory or workload
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/455Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
    • G06F9/45533Hypervisors; Virtual machine monitors
    • G06F9/45558Hypervisor-specific management and integration aspects
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/50Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5061Partitioning or combining of resources
    • G06F9/5072Grid computing
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00Administration; Management
    • G06Q10/06Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/50Network service management, e.g. ensuring proper service fulfilment according to agreements
    • H04L41/5041Network service management, e.g. ensuring proper service fulfilment according to agreements characterised by the time relationship between creation and deployment of a service
    • H04L41/5054Automatic deployment of services triggered by the service manager, e.g. service implementation by automatic configuration of network components
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L63/00Network architectures or network communication protocols for network security
    • H04L63/10Network architectures or network communication protocols for network security for controlling access to devices or network resources
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • H04L67/1001Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • H04L67/1001Protocols in which an application is distributed across nodes in the network for accessing one among a plurality of replicated servers
    • H04L67/1034Reaction to server failures by a load balancer
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/133Protocols for remote procedure calls [RPC]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/08Configuration management of networks or network elements
    • H04L41/0803Configuration setting
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04SSYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S40/00Systems for electrical power generation, transmission, distribution or end-user application management characterised by the use of communication or information technologies, or communication or information technology specific aspects supporting them

Definitions

  • the present invention relates to the field of cloud computing, and more particularly to a component-oriented hybrid cloud operating system architecture and a communication method thereof.
  • COS Cloud Operating System
  • the interface application and the intermediate functions of the management hardware combine a large number of heterogeneous devices into a logical resource pool, dynamically dispatch to the cloud application, and complete the service to the terminal.
  • the cloud data center environment is characterized by dynamic, heterogeneous, large-scale and single-point failure. Therefore, COS needs to adopt a widely compatible open architecture, considering compatibility with third-party hardware and software, and including secondary development.
  • COS will adopt scalability and high-availability design to achieve the scale expansion and business continuity goals pursued by cloud computing.
  • the single module architecture of the traditional OS can achieve efficient call between COS modules, but the coupling is tight, the structure is complex, and the system is difficult to expand.
  • the current cloud operating system lacks self-containedness, cannot describe and manage component components, and cannot dynamically monitor its processing environment.
  • the existing communication protocol is based on the stateless assumption design of the component, lacks the read-write separation mechanism and the load balancing strategy, and cannot achieve high availability and high-performance support for the stateful component processing cluster.
  • the efficiency of existing tree-based routing algorithms is affected by the expansion of the keyword size, and the hash-based routing algorithm causes a large amount of data movement when nodes change, and lacks A method for distributing data in a heterogeneous node load.
  • the object of the present invention is to provide a component-oriented hybrid cloud operating system architecture and communication method thereof.
  • the object of the present invention is to realize a hybrid architecture based on hierarchy, object and message model, and manage component components and their processing environment by using component-oriented ideas.
  • efficient routing of component processing clusters read-write separation and load balancing, to meet the open compatibility, loose coupling and scalability requirements of cloud operating systems, to solve the self-management problems of existing cloud operating systems, component horizontal scaling problems and high availability of stateful components.
  • the system is divided into portal layer, logic layer, adaptation layer and implementation layer from top to bottom.
  • Each layer is relatively independent. By defining standard interfaces in each layer to enhance openness, it is realized by adapting different functions in each layer. Enhance compatibility;
  • the cloud operating system is composed of a cloud portal, a cloud management portal, a cloud resource management, a monitoring management, a metering and billing, a service approval, and an authorization authentication function module, and each functional component communicates through a call based on a Rest message, which is free.
  • monitoring, billing, approval or other modules are customized and extended on demand, through the call based on the Rest message, on-demand distributed deployment and value-added development, Interoperability between different modules of the logic layer to enhance the scalability of the system; Introduce message-based communication mode to support asynchronous calls, use the message communication interface JMS to transmit the Rest message, and further decouple the system architecture.
  • the component management portal is responsible for managing the metadata of the component and monitoring its operational status;
  • the object architecture can realize on-demand expansion and distributed deployment, it belongs to RPC (Remote Process Call) synchronous communication mode, and the sender waits for the receiver to return before continuing execution.
  • the two processes are tightly coupled, and with the expansion and complication of the system, The association between components is too complicated.
  • a message-based communication method is introduced on the basis of the object architecture, and the message communication interface JMS is used to transmit the Rest information, so that the life cycle of the transmitting and receiving ends can be different, and asynchronous calling is supported.
  • the system architecture is further decoupled.
  • operations such as opening, closing, suspending, and shutting down the virtual machine between the portal and the cloud resource layer are implemented in an asynchronous manner, and the portal can return without issuing a command after issuing the command, and the promotion is improved.
  • the component management portal is responsible for managing the metadata information of the component, and supports operations such as registration, deletion, modification, and query.
  • the component is a triple including: name, service collection, access address, description;
  • the service is a four-tuple, including: name, type, message protocol, parameter list, key name, function description, non-functional description; in addition to describing and managing the component, the component management module further monitors its processing environment to ensure The scalability and availability of components provide basic services to improve the self-management capabilities of the cloud operating system.
  • the system assigns the user name user, password psw, and unique component id. After that, the access process of the component processing cluster is
  • the processing cluster initiates an access request to the system bus with the address url, and the system bus verifies the username, password, and id of the access node. If the verification is passed, the connection is established, and the code is:
  • Connection ConnectionFactory.createConnection(user,psw,url);
  • the component implements write processing in the onMessage method of writeTopicListener and registers with the system bus; Write— topic— consumer. setMessageListener(writeTopicListener);
  • each processing node corresponds to a queue subscription read operation
  • Read_queue session.createMultiQueue(num,id+”READ_QUEUE");
  • Read_queue_consumer session.createConsumer(read_queue);
  • Read_Queue_ consumer setMessageListener(readQueueListener);
  • the service type of each component is distinguished in the component management module, and it is set to idempotent or non-power, and the idempotent operation belongs to none.
  • State operation the result of each execution is the same in the same state; non-idempotent operations are stateful operations, and the results of each execution are different in the same state, the router routes according to the service type, and the non-power operation is sent to the unique write queue.
  • the power operation is sent to different read queues according to the load balancing policy; the read operation load balancing process is
  • the write operation is performed by pipeline method to improve the data writing efficiency.
  • the process is that node 1 first writes data, writes a data slice of 64 KB, and continues to receive data. Forwarding to node 2 at the same time 64K data written, node 2 to node n receive and forward data in the same manner until node n writes the last data packet that does not exceed 64 KB; based on the above communication method, the node monitoring module further processes the node processing node
  • the join, exit, invalidation, and recovery events are sent to the component management module, wherein the node join event refers to adding a processing node to the component; the node exit event refers to undoing a processing node for the component; the node invalid event refers to a processing node of the component Not available;
  • the node wakeup event refers to a component that is unavailable for processing node recovery available.
  • the invention also provides a component-oriented hybrid cloud operating system communication method, comprising a high-availability component cluster communication method and a horizontal flexible component cluster communication method, wherein: the high-availability component cluster communication method is a component management module for each component The cluster establishes a read-write separation queue group, and the component management module further divides the service type of the component into idempotent or non-power, the router performs routing according to the service type, the non-power operation is sent to the unique write queue, and the power operation is sent according to the load balancing policy.
  • the high-availability component cluster communication method is a component management module for each component
  • the cluster establishes a read-write separation queue group, and the component management module further divides the service type of the component into idempotent or non-power
  • the router performs routing according to the service type
  • the non-power operation is sent to the unique write queue
  • the power operation is sent according to the load balancing policy.
  • the node monitoring module sends the processing node join, exit, invalidation and recovery events to the component management module, and the component management module further adjusts the queue structure according to the node change event, and the method separates by reading and writing.
  • the load balancing improves the communication performance of the stateful component cluster, and adjusts the high availability of the queue structure guarantee component according to the node change
  • the component management module further adjusts the queue structure according to the node change event sent by the monitoring module, and the flow is as follows:
  • the horizontal telescopic component cluster communication method is that the component management module establishes a polygon queue structure for each service of the component number according to the number of nodes, and performs binary search routing according to the hash value, and the algorithm efficiency is determined by the number of hash packets far smaller than the keyword size, thereby improving the routing. Efficiency, and when the number of nodes changes, it is only necessary to adjust the data state of some nodes to improve the dynamic scaling efficiency.
  • the Hash interval to be processed by each node is divided according to the processing power of the node in the initialization phase.
  • the routing table realizes the proportional distribution of the data state to achieve load balancing; the component management module establishes a polygon queue group for each service of the component according to the number of nodes, each queue corresponds to a hash value interval or group, and performs a binary search route according to the hash value.
  • the routing algorithm is as follows:
  • the node where the interval is located is routed; the above algorithm is equivalent to performing binary search by grouping, and the complexity is 0 (log 2 n).
  • the efficiency of the algorithm is determined by the number of packets. Since the number of packets is much smaller than the keyword size, Therefore, the routing efficiency is improved.
  • the grouping strategy is adopted, when the number of nodes changes, only the key size of the partial group and the data state of the corresponding node need to be adjusted, thereby improving the dynamic scaling efficiency.
  • the beneficial effects of the present invention are: Compared with the prior art, the component-oriented hybrid architecture proposed by the present invention completes the open, compatible, extensible, loosely coupled componentized cloud operating system architecture, and through the component horizontally scalable cluster and The communication method of the highly available cluster guarantees the scalability and high availability of the cloud operating system.
  • FIG. 1 is a component-oriented hybrid COS architecture diagram
  • FIG. 2 is a communication architecture diagram supporting a highly available cluster of stateful components
  • 3 is a schematic diagram of a read/write separation queue group
  • FIG. 4 is a data write timing diagram
  • FIG. 5 is a node state transition diagram
  • FIG. 6 is a queue structure adjustment flowchart
  • FIG. 7 is a horizontal telescopic component cluster communication architecture diagram
  • Figure 9 is a flow chart of routing algorithm for multilateral queue groups.
  • Component-oriented hybrid cloud operating system architecture From the perspective of hierarchical model, the cloud operating system is divided into four levels: portal layer, logical layer, adaptation layer and implementation layer from top to bottom, as shown in Figure 1.
  • the layers are relatively independent. By defining standard interfaces at each layer to enhance the openness of the system, compatibility is enhanced by adapting different functions at each layer.
  • the separation of the user UI and the function logic can be achieved by stripping the portal layer.
  • the logic layer can support the secondary development of the portal or the third party by providing a unified standard Rest API; Compatible with a variety of specific functional implementations, for example, the cloud resource management module supports multiple virtualization infrastructures through virtualization adapters.
  • the monitoring management module is compatible with multiple monitoring protocols through the adaptation framework, and the process management module supports different approval processes through the process engine.
  • the customization, billing management and authentication authorization modules support plug-in access by reserving the hook interface.
  • the cloud operating system is composed of functional modules such as cloud portal, cloud management portal, cloud resource management, monitoring management, metering and billing, business approval, and authorization authentication. Each functional component communicates through a call based on the Rest message. Freely combined, distributed on demand, and can be developed with new modules based on demand, enhancing platform scalability by enabling interoperability between different modules in the logical layer.
  • the minimally installed cloud operating system is composed only of a cloud portal, a cloud management portal, and a cloud resource management module. On this basis, monitoring, billing, approval, or other modules can be customized and expanded as needed.
  • the object architecture can realize on-demand expansion and distributed deployment, it belongs to RPC (Remote Process Call) synchronous communication mode, and the sender waits for the receiver to return before continuing execution.
  • the two processes are tightly coupled, and the system is enlarged and complicated. The relationship between components is too complicated.
  • a message-based communication method is introduced on the basis of the object architecture, and the message communication interface JMS is used to transmit the Rest information, so that the life cycle of the transmitting and receiving ends can be different, and asynchronous calling is supported, so that the system architecture is further decoupled.
  • the operations of opening, closing, suspending, and shutting down the virtual machine between the portal and the cloud resource layer are implemented in an asynchronous manner, and the portal can return without issuing a response after issuing the command, thereby improving the user interaction effect.
  • the cloud operating system based on the above hybrid architecture can meet the requirements of open compatibility and expansion.
  • the component management portal is responsible for managing the metadata information of the component, and supports operations such as registration, deletion, modification, and query.
  • the component is a triplet ⁇ name, service collection, access address, Description ⁇
  • the service is a four-tuple ⁇ name, type, message protocol, parameter list, key name, function description, non-functional description ⁇ .
  • the component management module further monitors its processing environment to protect the component. Scalability and availability provide basic services to improve the self-management capabilities of the cloud operating system.
  • the present invention proposes an implementation of a cluster of high-availability components, as shown in FIG. 2, which mainly includes the following module component management module responsible for establishing, deleting, and adjusting a message queue group of components according to component information; the message router is responsible for sending a message to the queue component according to the routing information. .
  • the node monitoring module is responsible for detecting the joining and exiting of each node of the component processing cluster, and obtaining the resource configuration information of the processing node.
  • the component processing cluster is responsible for implementing the specific service functions of the component, and is composed of several processing nodes.
  • the component client is responsible for initiating a request to use the component service.
  • the system assigns the user name user, password psw, and only The component id, after which the access process of the component processing cluster is
  • the processing cluster initiates an access request to the system bus with the address url, and the system bus verifies the username, password, and id of the access node, and if the verification passes, establishes a connection.
  • the code is:
  • Connection ConnectionFactory.createConnection(user,psw,url);
  • the component implements write processing in the onMessage method of writeTopicListener and registers with the system bus;
  • each processing node corresponds to a queue order reading operation
  • Read_queue session.createMultiQueue(num,id+”READ_QUEUE");
  • Read_queue_consumer session.createConsumer(read_queue);
  • Read_ queue_ consumer setMessageListener(readQueueListener);
  • the service type of each component is distinguished in the component management module, and it is set to be idempotent or non-idempotent.
  • Idempotent operations are stateless operations, and the results of each execution are the same in the same state; non-idempotent operations are stateful operations, and the results of each execution in the same state are different.
  • the router routes according to the service type, and the non-idempotent operation is sent to the unique write queue.
  • the idempotent operation is sent to different read queues according to the load balancing policy, as shown in Figure 3.
  • the read operation load balancing process is
  • the node monitoring module further sends the component processing node join, exit, invalidation and recovery events to the component management module, and the node state transition relationship is as shown in FIG. 5.
  • the node join event refers to adding a processing node to the component; the node exit event refers to undoing a processing node for the component; the node invalidation event means that one processing node of the component is unavailable; the node waking event refers to an unusable processing node of the component Recovery is available.
  • the component management module further adjusts the queue structure according to the node change event sent by the monitoring module, and the process is as shown in FIG. 6:
  • a read operation queue is established for the node in the queue group, the node subscribes to the queue, and writes a topic to the write topic subscription;
  • the load balancing method of the stateless cluster distributes the operation to each node in a completely balanced manner, which will result in inconsistent processing results in the stateful state.
  • the above method establishes a read-write separation queue and targets the node Capability load balancing can avoid this problem, improve the communication performance of the stateful component cluster, and adjust the queue structure according to the node change event to ensure the high availability of the stateful cluster.
  • the embodiment of the invention provides a horizontal telescopic component cluster communication method, as shown in FIG. 7 .
  • the component management module establishes a polygon queue group for each service of the component according to the number of nodes, and each queue corresponds to a Hash value interval (packet), and performs a binary search route according to the hash value.
  • the routing algorithm is as shown in FIG. 9:
  • Initialization group lower limit i is 1
  • group upper limit j is the total number of groups m
  • the above algorithm is equivalent to performing binary search by grouping as a node, and the complexity is 0 (log 2 n).
  • the efficiency of the algorithm is determined by the number of packets. Since the number of packets is much smaller than the keyword size, the routing efficiency is improved.
  • the grouping strategy since the grouping strategy is adopted, when the number of nodes changes, it is only necessary to adjust the keyword size of the partial group and the data state of the corresponding node, thereby improving the dynamic scaling efficiency.
  • the present invention proposes a data state distribution method.
  • the Hash interval to be processed by each node is divided according to the node processing capability, and the calculation method of the node processing capability is as described above.
  • the method can divide the size of the Hash interval to be processed by the queue of each node according to the processing capability of the node, and realize the proportional distribution of the data state to achieve load balancing.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Theoretical Computer Science (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Computer Hardware Design (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Strategic Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Mathematical Physics (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Development Economics (AREA)
  • Computing Systems (AREA)
  • Computer Security & Cryptography (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Environmental & Geological Engineering (AREA)

Abstract

本发明提供一种面向构件的混合型云操作***体系结构及其通信方法,基于层次、对象和消息模型建立混合型架构,并采用面向构件思想对组成构件及其处理环境进行管理,在此基础上,对构件处理集群进行高效路由、读写分离和负载均衡,满足对云操作***地的开放兼容、松耦合和可扩展需求,解决现有云操作***的自管理问题、构件水平伸缩问题和有状态构件的高可用问题,与现有技术相比,本发明提出的面向构件的混合型架构完善了开放兼容、可扩展、松耦合的构件化云操作***体系结构,并通过构件水平伸缩集群和高可用集群的通信方法保障了云操作***的可伸缩性和高可用性。

Description

一种面向构件的混合型云操作***体系结构及其通信方法 技术领域
本发明涉及云计算领域, 具体地说是一种面向构件的混合型云操作***体 系结构及其通信方法。
背景技术
随着云计算的兴起, 传统的数据中心迅速向云数据中心转型。 在数据中心 从初级向高级形态演进的物理资源整合、 应用与虚拟化接合、 自动化管理、 数 据中心协同四个阶段中, 云操作***(COS, Cloud Operating System)发挥着重 要作用,承担着对上接口应用、 对下管理硬件的中间功能, 将大量的异构设备融 合为逻辑资源池, 动态调度给云应用, 完成对终端的服务。 云数据中心环境具有动态、 异构、 大规模和单点易失效的特征, 因此, COS 需采用广泛兼容的开放架构, 既考虑对第三方软硬件的兼容性, 也将二次开发 纳入其中, 提供完善标准的接口 API; 针对云计算环境对功能的动态变化需求, cos 需采用可扩展的构件化设计, 在虚拟化、 资源调度等基本构件基础上, 便 于运维管理、 计量计费和自助服务等构件的增值开发和按需部署; 此外, COS 还要采用伸缩性和高可用设计, 达到云计算追求的规模扩展和业务连续性目标。 针对云数据中心对 COS的松耦合、 可扩展、 可伸缩和高可用需求, 采用传 统 OS的单一模块架构可以实现 COS模块间的高效调用, 但耦合紧密、 结构复 杂, ***难于扩展; 采用层次架构可以使各模块间的组织结构和依赖关系清晰 化, 提高 COS的可靠性、 可移植性和维护性, 但软件栈层次太深使得内核过于 庞大, 并且模块间的耦合程度仍然较高, 不适于构建分布处理环境; 在上述架 构基础上, 开源软件 OpenStack和 CloudStack基于消息队列建立了松耦合的云 管理架构, 但缺乏面向构件的设计, 无法控制构件生命周期, 需要组成模块自 行考虑伸缩和高可用方式, 加重了模块的开发部署负担和运行开销。 遵循高内 聚低耦合原则, 应该从 COS层面增加对构件的管理并保障其可扩展性、 可伸缩 性和高可用性, 这其中主要面对的问题是
1.当前的云操作***缺乏自包含性,无法对组成构件进行描述和管理, 也无 法对其处理环境进行动态监控。 2.针对构件的高可用处理集群, 现有通信协议基于构件的无状态性假设设 计, 缺乏读写分离机制和负载均衡策略, 无法实现对有状态构件处理集群的高 可用和高性能支持。
3.针对构件的水平伸缩(scale-out) 处理集群, 现有基于树的路由算法效率 受到关键字规模扩大的影响,而基于 Hash 的路由算法在节点变化时会造成大量 的数据移动, 并且缺乏均衡异构节点负载的数据分布方法。
因此, 如何在 COS中提供对构件及其处理集群的管理和监控机制, 以及实 现消息的高效路由和负载均衡, 成为 COS架构中亟待解决的技术问题。
发明内容
本发明的目的是提供一种面向构件的混合型云操作***体系结构及其通信 方法。 本发明的目的是按以下方式实现的, 基于层次、 对象和消息模型建立混合 型架构, 并采用面向构件思想对组成构件及其处理环境进行管理, 在此基础上, 对构件处理集群进行高效路由、 读写分离和负载均衡, 满足对云操作***地的 开放兼容、 松耦合和可扩展需求, 解决现有云操作***的自管理问题、 构件水 平伸缩问题和有状态构件的高可用问题, 从层次模型角度, ***自上至下分为门户层、 逻辑层、 适配层和实现层, 各层相对独立, 通过在各层分别定义标准接口增强开放性, 通过在各层适配不 同功能实现增强兼容性;
从对象模型角度, 云操作***由云门户、 云管理门户、 云资源管理、 监控 管理、计量计费、业务审批、授权认证功能模块组成, 各功能组件通过基于 Rest 消息的调用进行通信, 可自由组合, 按需分布部署, 并且可根据需求增值开发 新模块, 通过在逻辑层的不同模块间实现互操作增强平台的可扩展性, 在实施 例中, 最小化安装的云操作***仅由云门户、 云管理门户和云资源管理模块组 成, 在此基础上, 监控、 计费、 审批或其他模块按需定制和扩展, 通过基于 Rest 消息的调用进行通信, 按需分布部署和增值开发, 通过在逻辑层的不同模块间 实现互操作增强***的可扩展性; 引入基于消息的通信方式支持异步调用, 使 用消息通信接口 JMS传输 Rest消息, 使***架构进一步解耦, 在此基础上, 采 用面向构建设计, 构件管理门户负责管理构件的元数据并对其运行状态进行监 控;
对象架构虽然能够实现按需扩展和分布部署, 但属于 RPC (Remote Process Call)同步通信方式,发送端等待接收端返回后才能继续执行,双方进程紧耦合, 随着***的扩大化和复杂化, 构件之间的关联关系过于复杂, 针对此问题, 在 对象架构的基础上引入基于消息的通信方式, 使用消息通信接口 JMS传输 Rest 信息, 使得发送和接收端生命周期可以不同, 并支持异步调用, 使***架构进 一步解耦, 在实施例中, 门户和云资源层之间的虚拟机开启、 关闭、 挂起、 关 闭等操作通过异步方式实现, 门户发出命令后无需等待响应即可返回, 提升了 用户交互效果;
基于上述混合型架构的云操作***能够满足开放兼容和扩展需求, 在此基 础上, 基于面向构建设计思想, 构件管理门户负责管理构件的元数据信息, 支 持注册、 删除、 修改和查询等操作, 其中, 构件是一个三元组包括: 名称、 服务集合、 访问地址、 描述;
服务是一个四元组, 包括: 名称、 类型、 消息协议、 参数列表、 key名称、 功能描述、 非功能描述; 除对构件进行描述和管理, 构件管理模块进一步对其处理环境进行监控, 为保障构件的可伸缩性和可用性提供基础服务, 完善云操作***的自管理能力, 在构件注册时,***为其分配用户名 user、 密码 psw以及唯一的构件 id,之后构件 处理集群的接入过程为
1 ) 处理集群向地址为 url的***总线发起接入请求, ***总线验证接入节 点的用户名、 密码和 id, 如验证通过, 建立连接, 代码为:
connection=ConnectionFactory.createConnection(user,psw,url);
2) 建立一个写操作主题, 构件的每个处理节点向该主题订阅写操作; write_topic=session.createTopic(id+"WRITE_TOPIC");
write_topic_consumer=session.createConsumer(write_topic);
3 )构件在 writeTopicListener的 onMessage方法中实现写处理, 并向***总 线注册; write— topic— consumer. setMessageListener(writeTopicListener);
4) 根据构件处理节点数 mum建立读操作队列组, 每个处理节点对应一个 队列订阅读操作
read_queue=session.createMultiQueue(num,id+"READ_QUEUE");
read_queue_consumer=session.createConsumer(read_queue);
5 )在 readQueueListener的 onMessage方法中实现具体的读处理功能, 并向 ***总线注册
read— queue— consumer. setMessageListener(readQueueListener); 在上述读写分离队列组基础上, 在构件管理模块中对各构件的服务类型进 行区分, 设定其为幂等或非幂, 幂等操作属于无状态操作, 在同一状态下每次 执行的结果相同; 非幂等操作属于有状态操作, 同一状态下每次执行的结果不 同, 路由器根据服务类型进行路由, 非幂操作发送到唯一的写队列, 幂操作根 据负载均衡策略发送到不同的读队列; 读操作负载均衡流程过程为
1 ) 计算节点的处理能力;
若节点 i的 CPU频率、 内存容量和 I/O带宽分别为 d, Μ^Ρ Β 集群的各 种资源为节点各种资源之和, 即 C=∑d, M=∑Mi, B=∑B1; 则节点 i 的 CPU权值为
Figure imgf000006_0001
I/O 带宽
Figure imgf000006_0002
若构件服务所需资源比例分别为 pepu
Figure imgf000006_0003
pro, 则节点 i的处理能力为
Figure imgf000006_0004
2) 根据读写操作权值计算各节点的负载;
若读队列! ^同写队列 Lw 的读写操作开销比 a ,则节点 i的负载
Figure imgf000006_0005
各节点的负载状态 Sfl Wi
3 ) 选择负载最轻的节点进行路由; 写操作采用流水方式进行, 以提高数据写入效率, 其过程为节点 1 首先写 入数据, 写入一个数据分片 64 KB后, 在继续接收数据的同时向节点 2转发己 写入的 64K数据, 节点 2至节点 n以相同方式接收和转发数据, 直到节点 n写 入最后一个不超过 64 KB数据分片; 在上述通信方法基础上, 节点监控模块进一步将构件处理节点的加入、 退 出、 失效和恢复事件发送给构件管理模块, 其中, 节点加入事件是指为构件添 加一个处理节点; 节点退出事件是指为构件撤销一个处理节点; 节点失效事件 是指构件的一个处理节点不可用; 节点苏醒事件是指构件一个不可用处理节点 恢复可用。 本发明还提供一种面向构件的混合型云操作***的通信方法, 包括高可用 构件集群通信方法和水平伸缩构件集群通信方法, 其中: 高可用构件集群通信方法, 是构件管理模块为每个构件集群建立一个读写 分离队列组, 构件管理模块进一步将构件的服务类型区分为幂等或非幂, 路由 器根据服务类型进行路由, 非幂操作发送到唯一的写队列, 幂操作根据负载均 衡策略发送到不同的读队列, 在此基础上, 节点监控模块将处理节点的加入、 退出、 失效和恢复事件发送给构件管理模块, 构件管理模块进一步根据节点变 化事件调整队列结构, 此方法通过读写分离和负载均衡提高有状态构件集群的 通信性能, 根据节点变化调整队列结构保障构件的高可用性, 构件管理模块进 一步根据监控模块发送的节点变化事件调整队列结构, 流程如下:
1 ) 当节点加入时, 在队列组中为该节点建立一个读操作队列, 该节点向该 队列订阅读操作, 并向写主题订阅写主题;
2) 当节点退出时, 删除该节点对应的读队列, 关闭其对写主题的订阅;
3 ) 当节点失效时, 停止向该节点的读队列发送读请求, 在写主题中为该节 点保留写操作;
4) 当节点苏醒时, 在写主题中同步写操作, 恢复向该节点对应的读队列发 送读操作请求; 无状态集群的负载均衡方法将操作完全均衡的分配到各个节点上, 在有状 态情况下将造成处理结果不一致, 上述方法通过建立读写分离队列并针对节点 能力负载均衡, 能够避免此问题, 提高有状态构件集群的通信性能, 并且根据 节点变化事件调整队列结构保障了有状态集群的高可用性; 水平伸缩构件集群通信方法, 是构件管理模块根据节点数为构件的每个服 务建立多边形队列结构, 按 Hash值进行二分查找路由, 算法效率由远小于关键 字规模的 Hash分组数目决定,提高了路由效率,并且在节点数发生变化的情况下 只需调整部分节点数据状态,提高了动态伸缩效率, 在上述通信方法基础上, 在 初始化阶段根据节点处理能力划分各节点所需处理的 Hash区间, 形成路由表, 实现按比例分布数据状态, 达到负载均衡; 构件管理模块根据节点数为构件的每个服务建立多边形队列组, 每个队列 对应一个 Hash值区间或分组, 按 Hash值进行二分查找路由, 路由算法如下:
1 ) 计算 key 的 Hash值 h;
2) 初始化分组下限 i为 1,分组上限 j为分组总数 m;
3 ) 重复;
4) 计算中间分组 t=(i+j)/2, 查看分组 t是否包含 h;
5 ) 若 h小于当前分组下限, 更新上限 j=t-l ;
6) 若 h大于当前分组上限, 更新下限1^+1;
7) 否则, 返回区间所在节点进行路由; 上述算法相当于以分组为节点做二分查找, 复杂度为 0(log2n),算法的效率 由分组数目决定,由于分组数目远小于关键字规模,因此提高了路由效率, 另外, 由于采用了分组策略,因此在节点数量发生变化的情况下只需调整部分分组的关 键字规模和对应节点的数据状态,提高了动态伸缩效率。 本发明的有益效果是: 与现有技术相比, 本发明提出的面向构件的混合型 架构完善了开放兼容、 可扩展、 松耦合的构件化云操作***体系结构, 并通过 构件水平伸缩集群和高可用集群的通信方法保障了云操作***的可伸缩性和高 可用性。
附图说明
图 1是面向构件的混合型 COS体系结构图; 图 2是支持有状态构件集群高可用的通信架构图; 图 3是读写分离队列组示意图; 图 4是数据写入时序图; 图 5是节点状态转换图; 图 6是队列结构调整流程图; 图 7 是水平伸缩构件集群通信架构图; 图 8 是多边队列组路由示意图; 图 9是多边队列组路由算法流程图。
具体实施方式
以下将结合附图及实施例来详细说明本发明的实施方式, 借此对本发明如 何应用技术手段来解决技术问题, 并达成技术效果的实现过程能充分理解并据 以实施。 需要说明的是, 如果不冲突, 本发明实施例以及实施例中的各个特征 的相互均在本发明的保护范围之内。
1.面向构件的混合型云操作***架构 从层次模型角度, 云操作***从上至下分为门户层、 逻辑层、 适配层和实 现层四个层次, 如图 1 所示。 各层相对独立, 通过在各层分别定义标准接口增 强***的开放性, 通过在各层适配不同功能实现增强兼容性。 在实施例中, 通 过剥离门户层可实现用户 UI和功能逻辑的分离, 例如, 逻辑层可通过对外提供 统一标准的 Rest API支持门户或第三方的二次开发; 通过抽象出逻辑功能层可 以实现对多种具体功能实现的兼容, 例如, 云资源管理模块通过虚拟化适配器 支持多种虚拟化基础设施, 监控管理模块通过适配框架兼容多种监控协议, 流 程管理模块通过流程引擎支持不同审批流程的定制, 计费管理和认证授权模块 通过预留钩子接口支持插件接入。 从对象模型角度, 云操作***由云门户、 云管理门户、 云资源管理、 监控 管理、 计量计费、 业务审批、 授权认证等功能模块组成, 各功能组件通过基于 Rest消息的调用进行通信, 可自由组合, 按需分布部署, 并且可根据需求增值 开发新模块, 通过在逻辑层的不同模块间实现互操作增强平台的可扩展性。 在 实施例中, 最小化安装的云操作***仅由云门户、 云管理门户和云资源管理模 块组成, 在此基础上, 监控、 计费、 审批或其他模块可以按需定制和扩展。
对象架构虽然可以实现按需扩展和分布部署, 但属于 RPC (Remote Process Call)同步通信方式,发送端等待接收端返回后才能继续执行,双方进程紧耦合, 随着***的扩大化和复杂化, 构件之间的关联关系过于复杂。 针对此问题, 在 对象架构的基础上引入基于消息的通信方式, 使用消息通信接口 JMS传输 Rest 信息, 使得发送和接收端生命周期可以不同, 并支持异步调用, 使***架构进 一步解耦。 在实施例中, 门户和云资源层之间的虚拟机开启、 关闭、 挂起、 关 闭等操作通过异步方式实现, 门户发出命令后无需等待响应即可返回, 提升了 用户交互效果。
基于上述混合型架构的云操作***能够满足开放兼容和扩展需求。 在此基 础上, 基于面向构建设计思想, 构件管理门户负责管理构件的元数据信息, 支 持注册、 删除、 修改和查询等操作, 其中, 构件是一个三元组 {名称、 服务集合、 访问地址、 描述 }
服务是一个四元组 {名称、 类型、 消息协议、 参数列表、 key名称、 功能描 述、 非功能描述 } 除对构件进行描述和管理, 构件管理模块进一步对其处理环境进行监控, 为保障构件的可伸缩性和可用性提供基础服务, 完善云操作***的自管理能力。
2.高可用构件集群通信方法
本发明提出高可用构件集群的实施例如图 2所示, 主要包括如下模块 构件管理模块负责根据构件信息建立、 删除和调整构件的消息队列组; 消息路由器负责根据路由信息向所述队列组分发消息。 节点监控模块负责检测构件处理集群各节点的加入和退出, 获取处理节点 的资源配置信息。
构件处理集群负责实现构件的具体服务功能, 由若干个处理节点组成。 构件客户端负责发起对构件服务的使用请求。
基于上述架构, 在构件注册时,***为其分配用户名 user、 密码 psw以及唯 一的构件 id,之后构件处理集群的接入过程为
1.处理集群向地址为 url的***总线发起接入请求, ***总线验证接入节点 的用户名、 密码和 id, 如验证通过, 建立连接。 代码为:
connection=ConnectionFactory.createConnection(user,psw,url);
2.建立一个写操作主题, 构件的每个处理节点向该主题订阅写操作; write_topic=session.createTopic(id+"WRITE_TOPIC");
write_topic_consumer=session.createConsumer(write_topic);
3.构件在 writeTopicListener的 onMessage方法中实现写处理, 并向***总 线注册;
write— topic— consumer. setMessageListener( writeTopicListener);
4.根据构件处理节点数 mum建立读操作队列组, 每个处理节点对应一个队 列订阅读操作
read_queue=session.createMultiQueue(num,id+"READ_QUEUE");
read_queue_consumer=session.createConsumer(read_queue);
5.在 readQueueListener的 onMessage方法中实现具体的读处理功能, 并向 ***总线注册
read— queue— consumer. setMessageListener(readQueueListener);
在上述读写分离队列组基础上, 在构件管理模块中对各构件的服务类型进 行区分, 设定其为幂等或非幂等。 幂等操作属于无状态操作, 在同一状态下每 次执行的结果相同; 非幂等操作属于有状态操作, 同一状态下每次执行的结果 不同。 路由器根据服务类型进行路由, 非幂等操作发送到唯一的写队列, 幂等 操作根据负载均衡策略发送到不同的读队列, 如图 3所示。 其中, 读操作负载均衡流程过程为
1.计算节点的处理能力。
若节点 i的 CPU频率、 内存容量和 I/O带宽分别为 d, ?^和^ , 集群的各 种资源为节点各种资源之和, 即 C=∑d, M=∑Mi, B=∑B10 则节点 i 的 CPU权值为
Figure imgf000011_0001
内存容量 W^^Mi/M, I/O 带宽
Figure imgf000012_0001
若构件服务所需资源比例分别为 pepu
Figure imgf000012_0002
pro, 则节点 i的处理能力为
Figure imgf000012_0003
2.根据读写操作权值计算各节点的负载。 若读队列! ^同写队列 Lw 的读写操作开销比 a
Figure imgf000012_0004
各节点的负载状态 S^L ^
3.选择负载最轻的节点进行路由。 写操作采用流水方式进行, 以提高数据写入效率, 如图 4所示, 其过程为 节点 1首先写入数据, 写入一个数据分片 (64 KB)后, 在继续接收数据的同时 向节点 2转发己写入的 64K数据,节点 2至节点 n以相同方式接收和转发数据, 直到节点 n写入最后一个数据分片 (不超过 64 KB)。 在上述通信方法基础上, 节点监控模块进一步将构件处理节点的加入、 退 出、 失效和恢复事件发送给构件管理模块, 节点状态转换关系如图 5所示。 其 中, 节点加入事件是指为构件添加一个处理节点; 节点退出事件是指为构件撤 销一个处理节点; 节点失效事件是指构件的一个处理节点不可用; 节点苏醒事 件是指构件一个不可用处理节点恢复可用。 构件管理模块进一步根据监控模块发送的节点变化事件调整队列结构, 流 程如图 6所示:
1.当节点加入时, 在队列组中为该节点建立一个读操作队列, 该节点向该 队列订阅读操作, 并向写主题订阅写主题;
2.当节点退出时, 删除该节点对应的读队列, 关闭其对写主题的订阅;
3.当节点失效时, 停止向该节点的读队列发送读请求, 在写主题中为该节 点保留写操作;
4.当节点苏醒时, 在写主题中同步写操作, 恢复向该节点对应的读队列发 送读操作请求。 无状态集群的负载均衡方法将操作完全均衡的分配到各个节点上, 在有状 态情况下将造成处理结果不一致。 上述方法通过建立读写分离队列并针对节点 能力负载均衡, 能够避免此问题, 提高有状态构件集群的通信性能, 并且根据 节点变化事件调整队列结构保障了有状态集群的高可用性。
3水平伸缩构件集群通信方法 本发明实施例提供了一种水平伸缩构件集群通信方法, 如图 7所示。 构件 管理模块根据节点数为构件的每个服务建立多边形队列组, 每个队列对应一个 Hash值区间 (分组), 按 Hash值进行二分查找路由, 路由算法为如图 9所示:
1.计算 key 的 Hash值 h
2.初始化分组下限 i为 1,分组上限 j为分组总数 m,
3.重复
4.计算中间分组 t=(i+j)/2, 查看分组 t是否包含 h
5.若 h小于当前分组下限, 更新上限 j=t- l
6.若 h大于当前分组上限, 更新下限1^+1
7.否则, 返回区间所在节点进行路由。 上述算法相当于以分组为节点做二分查找, 复杂度为 0(log2n),算法的效率 由分组数目决定,由于分组数目远小于关键字规模,因此提高了路由效率。 另外, 由于采用了分组策略,因此在节点数量发生变化的情况下只需调整部分分组的关 键字规模和对应节点的数据状态,提高了动态伸缩效率。
在上述通信方法基础上, 本发明实例提出一种数据状态分布方法, 在初始 化阶段根据节点处理能力划分各节点所需处理的 Hash区间, 节点处理能力的计 算方法如前所述。 此方法能够根据节点处理能力划分各节点的队列所需处理的 Hash区间规模, 实现按比例分布数据状态, 达到负载均衡。 本发明的其它特征和优点将在随后的说明书中阐述, 并且, 部分地从说明 书中变得显而易见, 或者通过实施本发明而了解。 本发明的目的和其他优点可 通过在说明书、 权利要求书以及附图中所特别指出的结构来实现和获得。

Claims

权利要求书
1、 一种面向构件的混合型云操作***, 其特征在于基于层次、对象和消息 模型建立混合型架构, 并采用面向构件思想对组成构件及其处理环境进行管理, 在此基础上, 对构件处理集群进行高效路由、 读写分离和负载均衡, 满足对云 操作***地的开放兼容、 松耦合和可扩展需求, 解决现有云操作***的自管理 问题、 构件水平伸缩问题和有状态构件的高可用问题, 从层次模型角度, ***自上至下分为门户层、 逻辑层、 适配层和实现层, 各层相对独立, 通过在各层分别定义标准接口增强开放性, 通过在各层适配不 同功能实现增强兼容性;
从对象模型角度, 云操作***由云门户、 云管理门户、 云资源管理、 监控 管理、计量计费、业务审批、授权认证功能模块组成, 各功能组件通过基于 Rest 消息的调用进行通信, 可自由组合, 按需分布部署, 并且可根据需求增值开发 新模块, 通过在逻辑层的不同模块间实现互操作增强平台的可扩展性, 在实施 例中, 最小化安装的云操作***仅由云门户、 云管理门户和云资源管理模块组 成, 在此基础上, 监控、 计费、 审批或其他模块按需定制和扩展, 通过基于 Rest 消息的调用进行通信, 按需分布部署和增值开发, 通过在逻辑层的不同模块间 实现互操作增强***的可扩展性; 引入基于消息的通信方式支持异步调用, 使 用消息通信接口 JMS传输 Rest消息, 使***架构进一步解耦, 在此基础上, 采 用面向构建设计, 构件管理门户负责管理构件的元数据并对其运行状态进行监 控;
对象架构虽然能够实现按需扩展和分布部署, 但属于 RPC (Remote Process Call)同步通信方式,发送端等待接收端返回后才能继续执行,双方进程紧耦合, 随着***的扩大化和复杂化, 构件之间的关联关系过于复杂, 针对此问题, 在 对象架构的基础上引入基于消息的通信方式, 使用消息通信接口 JMS传输 Rest 信息, 使得发送和接收端生命周期可以不同, 并支持异步调用, 使***架构进 一步解耦, 在实施例中, 门户和云资源层之间的虚拟机开启、 关闭、 挂起、 关 闭等操作通过异步方式实现, 门户发出命令后无需等待响应即可返回, 提升了 用户交互效果;
基于上述混合型架构的云操作***能够满足开放兼容和扩展需求, 在此基 础上, 基于面向构建设计思想, 构件管理门户负责管理构件的元数据信息, 支 持注册、 删除、 修改和查询等操作, 其中, 构件是一个三元组包括: 名称、 服务集合、 访问地址、 描述; 服务是一个四元组, 包括: 名称、 类型、 消息协议、 参数列表、 key名称、 功能描述、 非功能描述; 除对构件进行描述和管理, 构件管理模块进一步对其处理环境进行监控, 为保障构件的可伸缩性和可用性提供基础服务, 完善云操作***的自管理能力, 在构件注册时,***为其分配用户名 user、 密码 psw以及唯一的构件 id,之后构件 处理集群的接入过程为
1 ) 处理集群向地址为 url的***总线发起接入请求, ***总线验证接入节 点的用户名、 密码和 id, 如验证通过, 建立连接, 代码为:
connection=ConnectionFactory.createConnection(user,psw,url);
2) 建立一个写操作主题, 构件的每个处理节点向该主题订阅写操作; write_topic=session.createTopic(id+"WRITE_TOPIC");
write_topic_consumer=session.createConsumer(write_topic);
3 )构件在 writeTopicListener的 onMessage方法中实现写处理, 并向***总 线注册;
write— topic— consumer. setMessageListener( writeTopicListener);
4) 根据构件处理节点数 mum建立读操作队列组, 每个处理节点对应一个 队列订阅读操作
read_queue=session.createMultiQueue(num,id+"READ_QUEUE");
read_queue_consumer=session.createConsumer(read_queue);
5 )在 readQueueListener的 onMessage方法中实现具体的读处理功能, 并向 ***总线注册
read— queue— consumer. setMessageListener(readQueueListener);
在上述读写分离队列组基础上, 在构件管理模块中对各构件的服务类型进 行区分, 设定其为幂等或非幂, 幂等操作属于无状态操作, 在同一状态下每次 执行的结果相同; 非幂等操作属于有状态操作, 同一状态下每次执行的结果不 同, 路由器根据服务类型进行路由, 非幂操作发送到唯一的写队列, 幂操作根 据负载均衡策略发送到不同的读队列; 读操作负载均衡流程过程为
1 ) 计算节点的处理能力; 若节点 i的 CPU频率、 内存容量和 I/O带宽分别为 d Μ^Ρ Β 集群的各 种资源为节点各种资源之和, 即 C=∑d M=∑Mi B=∑B1; 则节点 i 的 CPU权值为 I/O 带宽
Figure imgf000016_0002
若构件服务所需资源比例分别为 pepu
Figure imgf000016_0003
pro, 则节点 i的处理能力为
Figure imgf000016_0004
2) 根据读写操作权值计算各节点的负载; 若读队列! ^同写队列 Lw 的读写操作开销比 a
Figure imgf000016_0005
各节点的负载状态 Sfl Wi
3 ) 选择负载最轻的节点进行路由; 写操作采用流水方式进行, 以提高数据写入效率, 其过程为节点 1 首先写 入数据, 写入一个数据分片 64 KB后, 在继续接收数据的同时向节点 2转发己 写入的 64K数据, 节点 2至节点 n以相同方式接收和转发数据, 直到节点 n写 入最后一个不超过 64 KB数据分片; 在上述通信方法基础上, 节点监控模块进一步将构件处理节点的加入、 退 出、 失效和恢复事件发送给构件管理模块, 其中, 节点加入事件是指为构件添 加一个处理节点; 节点退出事件是指为构件撤销一个处理节点; 节点失效事件 是指构件的一个处理节点不可用; 节点苏醒事件是指构件一个不可用处理节点 恢复可用。
2、 一种面向构件的混合型云操作***的通信方法, 其特征在于, 包括高可 用构件集群通信方法和水平伸缩构件集群通信方法, 其中: 高可用构件集群通信方法, 是构件管理模块为每个构件集群建立一个读写 分离队列组, 构件管理模块进一步将构件的服务类型区分为幂等或非幂, 路由 器根据服务类型进行路由, 非幂操作发送到唯一的写队列, 幂操作根据负载均 衡策略发送到不同的读队列, 在此基础上, 节点监控模块将处理节点的加入、 退出、 失效和恢复事件发送给构件管理模块, 构件管理模块进一步根据节点变 化事件调整队列结构, 此方法通过读写分离和负载均衡提高有状态构件集群的 通信性能, 根据节点变化调整队列结构保障构件的高可用性, 构件管理模块进 一步根据监控模块发送的节点变化事件调整队列结构, 流程如下:
1 ) 当节点加入时, 在队列组中为该节点建立一个读操作队列, 该节点向该 队列订阅读操作, 并向写主题订阅写主题;
2) 当节点退出时, 删除该节点对应的读队列, 关闭其对写主题的订阅;
3 ) 当节点失效时, 停止向该节点的读队列发送读请求, 在写主题中为该节 点保留写操作;
4) 当节点苏醒时, 在写主题中同步写操作, 恢复向该节点对应的读队列发 送读操作请求; 无状态集群的负载均衡方法将操作完全均衡的分配到各个节点上, 在有状 态情况下将造成处理结果不一致, 上述方法通过建立读写分离队列并针对节点 能力负载均衡, 能够避免此问题, 提高有状态构件集群的通信性能, 并且根据 节点变化事件调整队列结构保障了有状态集群的高可用性; 水平伸缩构件集群通信方法, 是构件管理模块根据节点数为构件的每个服 务建立多边形队列结构, 按 Hash值进行二分查找路由, 算法效率由远小于关键 字规模的 Hash分组数目决定,提高了路由效率,并且在节点数发生变化的情况下 只需调整部分节点数据状态,提高了动态伸缩效率, 在上述通信方法基础上, 在 初始化阶段根据节点处理能力划分各节点所需处理的 Hash区间, 形成路由表, 实现按比例分布数据状态, 达到负载均衡; 构件管理模块根据节点数为构件的每个服务建立多边形队列组, 每个队列 对应一个 Hash值区间或分组, 按 Hash值进行二分查找路由, 路由算法如下:
1 ) 计算 key 的 Hash值 h;
2) 初始化分组下限 i为 1,分组上限 j为分组总数 m; 3 ) 重复;
4) 计算中间分组 t=(i+j)/2, 查看分组 t是否包含 h;
5 ) 若 h小于当前分组下限, 更新上限 j=t-l ;
6) 若 h大于当前分组上限, 更新下限1^+1;
7) 否则, 返回区间所在节点进行路由;
上述算法相当于以分组为节点做二分查找, 复杂度为 0(log2n),算法的效率 由分组数目决定,由于分组数目远小于关键字规模,因此提高了路由效率, 另外, 由于采用了分组策略,因此在节点数量发生变化的情况下只需调整部分分组的关 键字规模和对应节点的数据状态,提高了动态伸缩效率。
PCT/CN2014/071233 2013-08-22 2014-01-23 一种面向构件的混合型云操作***体系结构及其通信方法 WO2015024368A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
JP2015543294A JP2015537307A (ja) 2013-08-22 2014-01-23 コンポーネント指向ハイブリッドクラウドオペレーティングシステムのアーキテクチャ及びその通信方法
EP14833104.4A EP2899947A4 (en) 2013-08-22 2014-01-23 ARCHITECTURE OF COMPRISING HYBRID CLOUD OPERATING SYSTEM, AND CORRESPONDING COMMUNICATION METHOD
US14/534,735 US9307017B2 (en) 2013-08-22 2014-11-06 Member-oriented hybrid cloud operating system architecture and communication method thereof

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310367864.2A CN103442049B (zh) 2013-08-22 2013-08-22 一种面向构件的混合型云操作***体系结构及其通信方法
CN201310367864.2 2013-08-22

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US14/534,735 Continuation US9307017B2 (en) 2013-08-22 2014-11-06 Member-oriented hybrid cloud operating system architecture and communication method thereof

Publications (1)

Publication Number Publication Date
WO2015024368A1 true WO2015024368A1 (zh) 2015-02-26

Family

ID=49695714

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/071233 WO2015024368A1 (zh) 2013-08-22 2014-01-23 一种面向构件的混合型云操作***体系结构及其通信方法

Country Status (5)

Country Link
US (1) US9307017B2 (zh)
EP (1) EP2899947A4 (zh)
JP (1) JP2015537307A (zh)
CN (1) CN103442049B (zh)
WO (1) WO2015024368A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110264467A (zh) * 2019-06-26 2019-09-20 西安电子科技大学 基于顶点切割的动态幂律图实时重划分方法
CN111277672A (zh) * 2020-03-31 2020-06-12 上海积成能源科技有限公司 一种基于非阻塞输入输出模型的能源物联网数据采集方法和软件网关
CN112118278A (zh) * 2019-06-04 2020-12-22 杭州海康威视***技术有限公司 计算节点接入方法、装置、电子设备及机器可读存储介质
CN115857885A (zh) * 2022-12-27 2023-03-28 深圳市浩科智联科技有限公司 一种基于soc软件的汽车主题ui全流程开发测试部署***

Families Citing this family (99)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103442049B (zh) 2013-08-22 2016-08-31 浪潮电子信息产业股份有限公司 一种面向构件的混合型云操作***体系结构及其通信方法
GB2521364A (en) * 2013-12-17 2015-06-24 Ibm Recording GUI data
US9420035B2 (en) * 2014-02-20 2016-08-16 International Business Machines Corporation Transaction isolation during multi-tenant transaction requests
CN103986662B (zh) * 2014-05-22 2017-12-05 浪潮电子信息产业股份有限公司 一种跨虚拟化平台的虚拟路由器实现方法
CN104135542B (zh) * 2014-08-19 2017-12-22 浪潮(北京)电子信息产业有限公司 一种云计算环境中异构负载均衡管理***和方法
US9729542B2 (en) 2014-09-24 2017-08-08 Oracle International Corporation Compartmentalizing application distribution for disparate electronic devices
US9501307B2 (en) 2014-09-26 2016-11-22 Comcast Cable Communications, Llc Systems and methods for providing availability to resources
CN104486440B (zh) * 2014-12-25 2018-03-02 浪潮电子信息产业股份有限公司 一种基于消息总线的云计算管理软件交互方法
US10193867B2 (en) 2015-05-27 2019-01-29 Ping Identity Corporation Methods and systems for API proxy based adaptive security
CN104980364A (zh) * 2015-06-23 2015-10-14 浪潮电子信息产业股份有限公司 一种基于智能查找算法的链路负载均衡方法
US10630649B2 (en) 2015-06-30 2020-04-21 K4Connect Inc. Home automation system including encrypted device connection based upon publicly accessible connection file and related methods
WO2017004200A1 (en) 2015-06-30 2017-01-05 K4Connect Inc. Home automation system including security controller for terminating communication with abnormally operating addressable devices and related methods
JP6739456B2 (ja) * 2015-06-30 2020-08-12 ケー4コネクト インコーポレイテッド クラウド及びホームメッセージキュー同期を含むホームオートメーションシステム、並びに関連方法
CN105338061B (zh) * 2015-09-29 2019-04-23 华中科技大学 一种轻量级消息中间件的实现方法与***
US10778809B2 (en) * 2016-02-26 2020-09-15 Arista Networks, Inc. Per-input port, per-control plane network data traffic class control plane policing
US9838376B1 (en) 2016-05-11 2017-12-05 Oracle International Corporation Microservices based multi-tenant identity and data security management cloud service
US10581820B2 (en) 2016-05-11 2020-03-03 Oracle International Corporation Key generation and rollover
US10425386B2 (en) 2016-05-11 2019-09-24 Oracle International Corporation Policy enforcement point for a multi-tenant identity and data security management cloud service
US10878079B2 (en) 2016-05-11 2020-12-29 Oracle International Corporation Identity cloud service authorization model with dynamic roles and scopes
US10341410B2 (en) 2016-05-11 2019-07-02 Oracle International Corporation Security tokens for a multi-tenant identity and data security management cloud service
US10454940B2 (en) 2016-05-11 2019-10-22 Oracle International Corporation Identity cloud service authorization model
US9838377B1 (en) 2016-05-11 2017-12-05 Oracle International Corporation Task segregation in a multi-tenant identity and data security management cloud service
US9781122B1 (en) 2016-05-11 2017-10-03 Oracle International Corporation Multi-tenant identity and data security management cloud service
US10735394B2 (en) 2016-08-05 2020-08-04 Oracle International Corporation Caching framework for a multi-tenant identity and data security management cloud service
US10263947B2 (en) 2016-08-05 2019-04-16 Oracle International Corporation LDAP to SCIM proxy service
US10516672B2 (en) 2016-08-05 2019-12-24 Oracle International Corporation Service discovery for a multi-tenant identity and data security management cloud service
US10505941B2 (en) 2016-08-05 2019-12-10 Oracle International Corporation Virtual directory system for LDAP to SCIM proxy service
US10530578B2 (en) 2016-08-05 2020-01-07 Oracle International Corporation Key store service
US10585682B2 (en) 2016-08-05 2020-03-10 Oracle International Corporation Tenant self-service troubleshooting for a multi-tenant identity and data security management cloud service
US10255061B2 (en) 2016-08-05 2019-04-09 Oracle International Corporation Zero down time upgrade for a multi-tenant identity and data security management cloud service
US10484382B2 (en) 2016-08-31 2019-11-19 Oracle International Corporation Data management for a multi-tenant identity cloud service
US10846390B2 (en) 2016-09-14 2020-11-24 Oracle International Corporation Single sign-on functionality for a multi-tenant identity and data security management cloud service
US10511589B2 (en) 2016-09-14 2019-12-17 Oracle International Corporation Single logout functionality for a multi-tenant identity and data security management cloud service
US10594684B2 (en) 2016-09-14 2020-03-17 Oracle International Corporation Generating derived credentials for a multi-tenant identity cloud service
US10567364B2 (en) 2016-09-16 2020-02-18 Oracle International Corporation Preserving LDAP hierarchy in a SCIM directory using special marker groups
US10791087B2 (en) 2016-09-16 2020-09-29 Oracle International Corporation SCIM to LDAP mapping using subtype attributes
US10341354B2 (en) 2016-09-16 2019-07-02 Oracle International Corporation Distributed high availability agent architecture
US10484243B2 (en) 2016-09-16 2019-11-19 Oracle International Corporation Application management for a multi-tenant identity cloud service
US10445395B2 (en) 2016-09-16 2019-10-15 Oracle International Corporation Cookie based state propagation for a multi-tenant identity cloud service
US10616224B2 (en) 2016-09-16 2020-04-07 Oracle International Corporation Tenant and service management for a multi-tenant identity and data security management cloud service
US10904074B2 (en) 2016-09-17 2021-01-26 Oracle International Corporation Composite event handler for a multi-tenant identity cloud service
US10182033B1 (en) * 2016-09-19 2019-01-15 Amazon Technologies, Inc. Integration of service scaling and service discovery systems
US10135916B1 (en) 2016-09-19 2018-11-20 Amazon Technologies, Inc. Integration of service scaling and external health checking systems
US10681012B2 (en) 2016-10-26 2020-06-09 Ping Identity Corporation Methods and systems for deep learning based API traffic security
CN106534306A (zh) * 2016-11-14 2017-03-22 北京大学(天津滨海)新代信息技术研究院 一种可扩展的异构云平台适配方法及其***
CN106657314B (zh) * 2016-12-14 2020-10-27 北京奇艺世纪科技有限公司 跨数据中心数据同步***及方法
US10261836B2 (en) * 2017-03-21 2019-04-16 Oracle International Corporation Dynamic dispatching of workloads spanning heterogeneous services
CN107220129B (zh) * 2017-04-14 2021-01-01 武汉斗鱼网络科技有限公司 一种软件模块间的通信方法和***
US10454915B2 (en) 2017-05-18 2019-10-22 Oracle International Corporation User authentication using kerberos with identity cloud service
US10310943B2 (en) * 2017-06-16 2019-06-04 Microsoft Technology Licensing, Llc Distributed data object management system
US10348858B2 (en) 2017-09-15 2019-07-09 Oracle International Corporation Dynamic message queues for a microservice based cloud service
CN107608804B (zh) * 2017-09-21 2020-06-12 浪潮云信息技术有限公司 一种任务处理***及方法
US10831789B2 (en) 2017-09-27 2020-11-10 Oracle International Corporation Reference attribute query processing for a multi-tenant cloud service
US11271969B2 (en) 2017-09-28 2022-03-08 Oracle International Corporation Rest-based declarative policy management
US10834137B2 (en) 2017-09-28 2020-11-10 Oracle International Corporation Rest-based declarative policy management
US10705823B2 (en) 2017-09-29 2020-07-07 Oracle International Corporation Application templates and upgrade framework for a multi-tenant identity cloud service
US10608951B2 (en) * 2017-09-30 2020-03-31 Oracle International Corporation Live resegmenting of partitions in distributed stream-processing platforms
US10699010B2 (en) 2017-10-13 2020-06-30 Ping Identity Corporation Methods and apparatus for analyzing sequences of application programming interface traffic to identify potential malicious actions
CN108200162A (zh) * 2017-12-30 2018-06-22 中建材信息技术股份有限公司 一种基于公有云实现的数据库可用互备方法
US10715564B2 (en) 2018-01-29 2020-07-14 Oracle International Corporation Dynamic client registration for an identity cloud service
US10931656B2 (en) 2018-03-27 2021-02-23 Oracle International Corporation Cross-region trust for a multi-tenant identity cloud service
US10798165B2 (en) 2018-04-02 2020-10-06 Oracle International Corporation Tenant data comparison for a multi-tenant identity cloud service
US11165634B2 (en) 2018-04-02 2021-11-02 Oracle International Corporation Data replication conflict detection and resolution for a multi-tenant identity cloud service
US11258775B2 (en) 2018-04-04 2022-02-22 Oracle International Corporation Local write for a multi-tenant identity cloud service
CN108769049A (zh) * 2018-06-08 2018-11-06 郑州云海信息技术有限公司 一种向openstack同步用户权限的方法和服务器
US11012444B2 (en) 2018-06-25 2021-05-18 Oracle International Corporation Declarative third party identity provider integration for a multi-tenant identity cloud service
US10764273B2 (en) 2018-06-28 2020-09-01 Oracle International Corporation Session synchronization across multiple devices in an identity cloud service
CN109324807B (zh) * 2018-09-27 2021-09-03 京信网络***股份有限公司 OpenStack的计费模块安装方法和***
US11693835B2 (en) 2018-10-17 2023-07-04 Oracle International Corporation Dynamic database schema allocation on tenant onboarding for a multi-tenant identity cloud service
US11321187B2 (en) 2018-10-19 2022-05-03 Oracle International Corporation Assured lazy rollback for a multi-tenant identity cloud service
CN109583832A (zh) * 2018-11-16 2019-04-05 上海浦东发展银行股份有限公司***中心 一种用于银行管理的工作流智能流转管理***
US20200175361A1 (en) * 2018-11-30 2020-06-04 Alibaba Group Holding Limited Partitioning of deep learning inference with dynamic offloading
CN109639799B (zh) * 2018-12-12 2021-08-17 北京锐安科技有限公司 异构云平台的处理方法及装置
EP3678348A1 (en) 2019-01-04 2020-07-08 Ping Identity Corporation Methods and systems for data traffic based adpative security
US11651357B2 (en) 2019-02-01 2023-05-16 Oracle International Corporation Multifactor authentication without a user footprint
US11061929B2 (en) 2019-02-08 2021-07-13 Oracle International Corporation Replication of resource type and schema metadata for a multi-tenant identity cloud service
US11321343B2 (en) 2019-02-19 2022-05-03 Oracle International Corporation Tenant replication bootstrap for a multi-tenant identity cloud service
US11669321B2 (en) 2019-02-20 2023-06-06 Oracle International Corporation Automated database upgrade for a multi-tenant identity cloud service
US11792226B2 (en) 2019-02-25 2023-10-17 Oracle International Corporation Automatic api document generation from scim metadata
US11423111B2 (en) 2019-02-25 2022-08-23 Oracle International Corporation Client API for rest based endpoints for a multi-tenant identify cloud service
US11687378B2 (en) 2019-09-13 2023-06-27 Oracle International Corporation Multi-tenant identity cloud service with on-premise authentication integration and bridge high availability
US11870770B2 (en) 2019-09-13 2024-01-09 Oracle International Corporation Multi-tenant identity cloud service with on-premise authentication integration
US11611548B2 (en) 2019-11-22 2023-03-21 Oracle International Corporation Bulk multifactor authentication enrollment
CN111726418B (zh) * 2020-06-30 2023-08-11 平安银行股份有限公司 云资源的调配方法、装置、终端设备及存储介质
CN112083914B (zh) * 2020-08-31 2023-09-12 深圳航天科技创新研究院 实现对象模型嵌入式操作***软总线的方法及***
CN112307128B (zh) * 2020-11-26 2024-07-16 中国—东盟信息港股份有限公司 一种基于分布式的异构数据同步***及方法
CN112488506B (zh) * 2020-11-30 2024-07-12 哈尔滨工程大学 一种智能无人***集群的可扩展分布式架构及自组织方法
EP4044031A4 (en) * 2020-12-16 2023-12-13 Kuyun (Shanghai) Information Technology Co., Ltd. CLOUD ORCHESTRATION SYSTEM AND METHOD BASED ON READ-WRITE SEPARATION AND AUTOMATIC SCALING
CN112738193B (zh) * 2020-12-24 2022-08-19 青岛民航凯亚***集成有限公司 云计算的负载均衡方法及装置
CN112685199B (zh) * 2020-12-30 2023-10-20 董小君 一种消息队列修复方法、装置、计算机设备及存储介质
CN113128909A (zh) * 2021-05-13 2021-07-16 瑞麟天下节能技术(北京)有限公司 一种基于电力资源的服务器管理方法
CN113423025B (zh) * 2021-06-22 2024-02-13 烟台东方智能技术有限公司 一种具有人工智能的数据管理终端
CN115412478B (zh) * 2022-08-30 2023-06-02 奥特酷智能科技(南京)有限公司 一种面向服务的汽车域控制器数据转发***及方法
CN115473898B (zh) * 2022-11-14 2023-03-24 杭州欧若数网科技有限公司 一种边缘计算场景下的图数据库管理方法和***
CN115733848B (zh) * 2022-11-16 2024-06-25 北京航空航天大学 一种面向边缘设备的数据分布式存储管理***
CN116522323B (zh) * 2023-03-17 2023-11-24 哈尔滨工业大学 一种基于命名空间的容器消息队列读写管理方法
CN117112231B (zh) * 2023-09-22 2024-04-16 中国人民解放军91977部队 一种多模型协同处理方法及装置
CN117290076B (zh) * 2023-11-24 2024-04-09 苏州盈数智能科技有限公司 一种易扩缩容的多并发任务派发的设备管理方法及***
CN117348852B (zh) * 2023-12-05 2024-03-29 华南理工大学 基于细粒度软件要素构建数据链路的方法、装置及介质

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521022A (zh) * 2010-10-27 2012-06-27 微软公司 在无状态云计算环境中操作的有状态的应用程序
CN102970332A (zh) * 2012-10-26 2013-03-13 江苏物联网研究发展中心 云os组件自主式组装***
CN103442049A (zh) * 2013-08-22 2013-12-11 浪潮电子信息产业股份有限公司 一种面向构件的混合型云操作***体系结构及其通信方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100521603C (zh) * 2004-07-13 2009-07-29 联想网御科技(北京)有限公司 集群模式下实现网络安全设备高可用性的方法
US8504774B2 (en) * 2010-10-13 2013-08-06 Microsoft Corporation Dynamic cache configuration using separate read and write caches
US9027024B2 (en) * 2012-05-09 2015-05-05 Rackspace Us, Inc. Market-based virtual machine allocation

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521022A (zh) * 2010-10-27 2012-06-27 微软公司 在无状态云计算环境中操作的有状态的应用程序
CN102970332A (zh) * 2012-10-26 2013-03-13 江苏物联网研究发展中心 云os组件自主式组装***
CN103442049A (zh) * 2013-08-22 2013-12-11 浪潮电子信息产业股份有限公司 一种面向构件的混合型云操作***体系结构及其通信方法

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
LI, MINGJIE: "INSPUR IN -CLOUD OS V2.0 PRODUCT ANALYSIS(VOLUME 1", SCIENCE & TECHNOLOGY INSPUR, vol. 1, no. 4, 30 April 2012 (2012-04-30), XP008176083 *
See also references of EP2899947A4 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112118278A (zh) * 2019-06-04 2020-12-22 杭州海康威视***技术有限公司 计算节点接入方法、装置、电子设备及机器可读存储介质
CN110264467A (zh) * 2019-06-26 2019-09-20 西安电子科技大学 基于顶点切割的动态幂律图实时重划分方法
CN110264467B (zh) * 2019-06-26 2022-12-06 西安电子科技大学 基于顶点切割的动态幂律图实时重划分方法
CN111277672A (zh) * 2020-03-31 2020-06-12 上海积成能源科技有限公司 一种基于非阻塞输入输出模型的能源物联网数据采集方法和软件网关
CN111277672B (zh) * 2020-03-31 2022-03-11 上海积成能源科技有限公司 一种基于非阻塞输入输出模型的能源物联网数据采集方法
CN115857885A (zh) * 2022-12-27 2023-03-28 深圳市浩科智联科技有限公司 一种基于soc软件的汽车主题ui全流程开发测试部署***
CN115857885B (zh) * 2022-12-27 2023-09-22 深圳市浩科智联科技有限公司 一种基于soc软件的汽车主题ui全流程开发测试部署***

Also Published As

Publication number Publication date
US9307017B2 (en) 2016-04-05
CN103442049B (zh) 2016-08-31
EP2899947A1 (en) 2015-07-29
US20150067135A1 (en) 2015-03-05
JP2015537307A (ja) 2015-12-24
CN103442049A (zh) 2013-12-11
EP2899947A4 (en) 2015-09-23

Similar Documents

Publication Publication Date Title
WO2015024368A1 (zh) 一种面向构件的混合型云操作***体系结构及其通信方法
US10223140B2 (en) System and method for network function virtualization resource management
WO2021213004A1 (zh) 一种微服务管理***、部署方法及相关设备
TWI479850B (zh) 單數據機板之改良式多胞元支援方法及系統
WO2018176998A1 (zh) 数据存储方法及装置
WO2018006676A1 (zh) 加速资源处理方法、装置及网络功能虚拟化***
EP2838243B1 (en) Capability aggregation and exposure method and system
US20120324460A1 (en) Thread Execution in a Computing Environment
WO2017049945A1 (zh) 加速器虚拟化的方法、装置及集中资源管理器
CN111857873A (zh) 一种实现云原生容器网络的方法
WO2013163865A1 (zh) 虚拟机热迁移和部署的方法、服务器及集群***
WO2021155667A1 (zh) 模型训练方法、装置及集群***
CN105187512A (zh) 一种虚拟机集群负载均衡方法及***
WO2015192584A1 (zh) 虚拟路由***及方法
Zou et al. Design and implementation of hybrid cloud computing architecture based on cloud bus
CN103716397B (zh) 一种面向服务仿真时钟推进方法
WO2022120806A1 (zh) 面向高性能计算多云的分布式消息传递方法及***
WO2022028092A1 (zh) 一种vnf实例化的方法和装置
WO2019153880A1 (zh) 集群中镜像文件下载的方法、节点、查询服务器
KR20210122849A (ko) 단계 지원 작업 흐름을 위한 방법 및 장치
Fei et al. FlexNFV: Flexible network service chaining with dynamic scaling
George et al. Hadoop MapReduce for mobile clouds
Hasan et al. Dynamic load balancing model based on server status (DLBS) for green computing
Huang et al. Blockchain based log system
CN103312808B (zh) 一种云操作***总线及通信方法

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2015543294

Country of ref document: JP

Kind code of ref document: A

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14833104

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE