-
公开(公告)号：US10785272B2
公开(公告)日：2020-09-22
申请号：US15492392
申请日：2017-04-20
Applicant: Huawei Technologies Co., Ltd.
Inventor： Sibao Hong , Mingzhen Xia , Songshan Zhang
Abstract: A streaming application upgrading method and a stream computing system, where the method includes obtaining a updated logical model of a streaming application, determining a to-be-adjusted stream by comparing the updated logical model with an initial logical model, generating an upgrading instruction according to the to-be-adjusted stream, and delivering the generated upgrading instruction to a worker node such that the worker node adjusts, according to an indication of the upgrading instruction, a stream between process elements (PEs) distributed on the worker node. The method provided in the present disclosure can upgrade the streaming application online without interrupting a service.
-
公开(公告)号：US11256749B2
公开(公告)日：2022-02-22
申请号：US16426287
申请日：2019-05-30
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
Inventor： Yinglong Xia , Jian Xu , Mingzhen Xia
IPC: G06F15/16 , G06F16/901 , G06F16/27 , G06F9/48 , G06F9/50
Abstract: A graph data processing method and a distributed system is disclosed. The distributed system includes a master node and a plurality of worker nodes. The master node obtains master node graph data, divides the graph data to obtain P shards, where the P shards include a first shard and a second shard. The master node determines at least two edge sets from each shard, schedules at least two edge sets included in the first shard to at least two worker nodes for processing, and schedules an associate edge set included in the second shard to the at least two worker nodes for processing, where the associate edge set is an edge set that includes an outgoing edge of a target vertex corresponding to the first shard.
-
公开(公告)号：US20190332944A1
公开(公告)日：2019-10-31
申请号：US16425012
申请日：2019-05-29
Applicant: Huawei Technologies Co., Ltd.
Inventor： Xiaolong Bai , Changzheng Zhang , Mingzhen Xia
Abstract: A training method, apparatus, and chip for a neural network model includes determining a model training mode of each layer based on an estimated data volume in a model parameter set and an estimated data volume of output data of the layer, obtaining second output data that is obtained by m worker modules by training a (j−1)th layer, and directly obtaining by a worker module a global gradient of a model parameter by training the model parameter based on the second output data when a model parallel training mode is used for a jth layer.
-
公开(公告)号：US20170300367A1
公开(公告)日：2017-10-19
申请号：US15640685
申请日：2017-07-03
Applicant: Huawei Technologies Co., Ltd.
Inventor： Songshan Zhang , Mingzhen Xia
CPC classification number: G06F9/5083 , G06F9/3005 , G06F16/24568 , G06F16/9024
Abstract: A streaming graph optimization method and apparatus are disclosed, relating to the stream processing field. A stream application streaming graph provided by a user is received and the streaming graph is parsed and a streaming graph described by an operator node and a data stream side is constructed. Additionally the streaming graph is disassembled according to a maximum atom division principle, so as to obtain at least one streaming subgraph and adjacency operator combination is performed on the at least one streaming subgraph according to a combination algorithm, so as to obtain an optimized streaming graph.
-
公开(公告)号：US12147473B2
公开(公告)日：2024-11-19
申请号：US17580605
申请日：2022-01-20
Applicant: HUAWEI TECHNOLOGIES CO., LTD.
Inventor： Yinglong Xia , Jian Xu , Mingzhen Xia
IPC: G06F15/16 , G06F9/48 , G06F9/50 , G06F16/27 , G06F16/901
Abstract: Graph data processing methods and system are disclosed. One example method comprises obtaining, by a master node, graph data, wherein the graph data comprises M vertexes and a plurality of directional edges, each edge connects two vertexes, a direction of each edge is from a source to a destination vertex in the two vertexes, and M is an integer greater than two. The node divides the graph data into P non-overlapping shards, where each shard comprises at least one incoming edge directed to at least one vertex in the corresponding shard. The node schedules at least two edge sets comprised in a first shard of the P shards and an associate edge set comprised in a second shard of the P shards for processing by at least two worker nodes.
-
Patent Agency Ranking
公开(公告)号：US11010681B2
公开(公告)日：2021-05-18
申请号：US16805007
申请日：2020-02-28
Applicant: Huawei Technologies Co., Ltd.
Inventor： Jian Lin , Mingzhen Xia
IPC: G06N7/00 , G06F16/901 , H04L12/803 , H04L12/851
Abstract: A distributed computing system is provided. Both a first computing node and a second computing node in the distributed computing system store information about a name, a size, and a communication peer side identifier of a first data flow graph parameter in a data flow graph. The first computing node stores the first data flow graph parameter, where the first computing node and the second computing node generate respective triplets based on same interface parameter generation algorithms and information about the first data flow graph parameter that are stored in the respective nodes. The triplet is used as an interface parameter of a message passing interface (MPI) primitive that is used to transmit the first data flow graph parameter between the first computing node and the second computing node.
-
7.发明申请公开(公告)号：US20180367584A1
公开(公告)日：2018-12-20
申请号：US16112236
申请日：2018-08-24
Applicant: HUAWEI TECHNOLOGIES CO.,LTD.
Inventor： Yunlong Shi , Mingzhen Xia , Cheng He
IPC: H04L29/06 , G06F9/50 , H04L12/911 , H04L12/26
Abstract: A stream computer system and a method for processing a data stream in a stream computing system are disclosed. The method includes a first working node invokes at least one execution unit to process a data stream according to an initial parallelism degree, a control node collects information reflecting data traffic between the first working node and a second working node, and information reflecting data processing speed of the first working node, determines an optimized parallelism degree for the first working node according to the collected information, and adjusts the parallelism degree of the first working node to be consistent with the optimized parallelism degree.
-
8.发明授权公开(公告)号：US10097595B2
公开(公告)日：2018-10-09
申请号：US15257722
申请日：2016-09-06
Applicant: Huawei Technologies Co., Ltd.
Inventor： Yunlong Shi , Mingzhen Xia , Cheng He
IPC: H04L29/06 , G06F9/50 , H04L12/26 , H04L12/911
Abstract: A stream computer system and a method for processing a data stream in a stream computing system are disclosed. In an embodiment, the method includes collecting data traffic information between each working node and other working nodes and processing speed information for each working node, determining an optimized parallelism degree for each working node according to the collected data traffic information and processing speed information and adjusting a parallelism degree of the working node according to the optimized parallelism degree of the working node.
-
公开(公告)号：US20180131608A1
公开(公告)日：2018-05-10
申请号：US15720171
申请日：2017-09-29
Applicant: HUAWEI TECHNOLOGIES CO.,LTD.
Inventor： Song Jiang , Mingzhen Xia
IPC: H04L12/741 , H04L12/801 , H04L12/717 , H04L12/715
Abstract: The present disclosure discloses a software defined network SDN-based data processing system, and the system includes: a source data node, configured to receive a first data packet, and send to a corresponding source control node; the source control node, configured to receive the first data packet, where the first data packet carries a destination address of the first data packet; and determine a destination control node; and the destination control node, configured to receive the first data packet, and generate a second data packet and a matching policy rule. According to a software defined network-based data processing system in an embodiment of the present disclosure, the collaboration capability between nodes is improved so as to reduce the redundancy of multi-node processing in a network device, thereby improving the service processing efficiency of the network. The present disclosure further discloses a software defined network-based data processing method and device.
-
10.发明授权公开(公告)号：US09203748B2
公开(公告)日：2015-12-01
申请号：US14096586
申请日：2013-12-04
Applicant: Huawei Technologies Co., Ltd.
Inventor： Song Jiang , Mingzhen Xia
IPC: H04L12/715 , H04L12/741 , H04L12/717 , H04L12/801
Abstract: The present disclosure discloses a software defined network SDN-based data processing system, and the system includes: a source data node, configured to receive a first data packet, and send to a corresponding source control node; the source control node, configured to receive the first data packet, where the first data packet carries a destination address of the first data packet; and determine a destination control node; and the destination control node, configured to receive the first data packet, and generate a second data packet and a matching policy rule. According to a software defined network-based data processing system in an embodiment of the present disclosure, the collaboration capability between nodes is improved so as to reduce the redundancy of multi-node processing in a network device, thereby improving the service processing efficiency of the network. The present disclosure further discloses a software defined network-based data processing method and device.
Abstract translation: 本公开公开了一种基于软件定义的基于网络SDN的数据处理系统，所述系统包括：源数据节点，被配置为接收第一数据分组，并发送到对应的源控制节点; 所述源控制节点被配置为接收所述第一数据分组，其中所述第一数据分组携带所述第一数据分组的目的地地址; 并确定目的地控制节点; 以及所述目的地控制节点，被配置为接收所述第一数据分组，并且生成第二数据分组和匹配策略规则。 根据本公开的实施例中的基于软件定义的基于网络的数据处理系统，提高了节点之间的协作能力，以减少网络设备中多节点处理的冗余度，从而提高了网络设备的服务处理效率 网络。 本公开还公开了一种基于软件的网络数据处理方法和装置。
-
-
-
-
-
-
-
-
-
Search Results
16
records
(took 0.025 seconds)
