Future Wei Technologies, Inc.

Alibaba Group Holding Limited

Huawei Technologies Co., Limited

Le Cai

Qingqing Zhou

Yang Sun

Jason Yang Sun

Guogen Zhang

Yu Cheng Chen

Cheng Zhu

Jianjun Chen

Kamini Jagtiani

Chunfeng Pei

Mason Sharp

Mingyi Zhang

Jun Chen

Demai Ni

Yongyan Wang

Xin Jia

Qinyi Wu

Yu Chen

Xiaolin Du

A method for data redistribution of a job data in a first datanode (DN) to at least one additional DN in a Massively Parallel Processing (MPP) Database (DB) is provided. The method includes recording a snapshot of the job data, creating a first data portion in the first DN and a redistribution data portion in the first DN, collecting changes to a job data copy stored in a temporary table, and initiating transfer of the redistribution data portion to the at least one additional DN.

System and method for data redistribution in a database

Transaction processing for a database distributed across availability zones is disclosed, including: determining that a transaction comprising a set of one or more statements is to be executed on a plurality of database servers across at least two availability zones, wherein each availability zone is associated with a respective hybrid logical clock (HLC)-based centralized time service; causing the set of one or more statements to execute on the plurality of database servers across the at least two availability zones; obtaining a plurality of HLC-based prepare timestamps from the plurality of database servers across the at least two availability zones; selecting a maximum HLC-based prepare timestamp to use as a commit timestamp associated with the transaction; and determining whether to return a commit result corresponding to the transaction after a time delay that is determined based at least in part on a predetermined time skew.

Transaction processing for a database distributed across availability zones

A computer-implemented method of relocating data in a distributed database comprises: creating, by one or more processors, a second table in the distributed database, the second table including all columns from a first table; copying, by the one or more processors, a first set of tuples from the first table to the second table; modifying, by the one or more processors, during the copying of the first set of tuples, data of the first table according to a modification; after the copying of the first set of tuples, modifying, by the one or more processors, data of the second table according to the modification; and switching, by the one or more processors, the second table for the first table in a catalog of the distributed database.

Redistributing table data in a database cluster

A system and method for improved online transaction processing (OLTP) in a sharded database is provided. Overhead associated with a global transaction manager is reduced and scalability improved by determining whether incoming queries are single-shard transactions or multi-shard transactions. For multi-shard transactions, a distributed transaction ID (DXID) is requested from the GTM, and then forwarded with the query to one or more data notes. For single-shard transactions, the query is sent to a data node without requesting a DXID from the GTM.

High-throughput distributed transaction management for globally consistent sharded OLTP system and method of implementing

A system (and associated method) includes a processor which determines a performance metric ratio of a performance metric of a second type of server relative to a performance metric of a first server type for each of multiple sizes of multiple benchmark datasets to thereby determine a plurality of performance metric ratio values. The processor also determines an interpolation function for the plurality of performance metric ratio values usable to compute interpolation performance metric ratios of the second type of server relative to the first type of server for dataset sizes other than the first plurality of sizes. Given a second dataset, the processor determines an amount of the second dataset to provide to each of the respective server types using the interpolation function. The processor configures a load balancer based on the amount of the second dataset determined for each type of server.

Balancing CPU utilization in a massively parallel processing heterogeneous cluster

A computer-implemented method for distributing a plurality of data partitions among a cluster of computer nodes includes dividing a database into a plurality of primary partitions, storing the primary partitions on a plurality of nodes associated with a computer network, duplicating the primary partitions to create a plurality of secondary partitions and storing the secondary partitions on the nodes, wherein the primary and secondary partitions are substantially evenly distributed among the nodes, each secondary partition is stored on a different node than its corresponding primary partition, diversity exists with respect to each secondary partition and the node where it is stored, the primary partitions are available to serve a system workload, and one or more secondary partitions corresponding to one or more primary partitions of a failed node are available to be upgraded to primary status to maintain workload balancing and data availability after one of the computer nodes fails.

Replicated database distribution for workload balancing after cluster reconfiguration

A method includes dividing a dataset into partitions by hashing a specified key, selecting a set of distributed file system nodes as a primary node group for storage of the partitions, and causing a primary copy of the partitions to be stored on the primary node group by a distributed storage system file server such that the location of each partition is known by hashing of the specified key.

Data placement control for distributed computing environment

In one embodiment, a method includes determining a number of initial servers in a massively parallel processing (MPP) database cluster and determining an initial bucket configuration of the MPP database cluster, where the initial bucket configuration has a number of initial buckets. The method also includes adding a number of additional servers to the MPP database cluster to produce a number of updated servers, where the updated servers include the initial servers and the additional servers and creating an updated bucket configuration in accordance with the number of initial servers, the initial bucket configuration, and the number of additional servers, where the updated bucket configuration has a number of updated buckets. Additionally, the method includes redistributing data of the MPP cluster in accordance with the updated bucket configuration.

System and method for massively parallel processor database

In one embodiment, a method of performing point-in-time recovery (PITR) in a massively parallel processing (MPP) database includes receiving, by a data node from a coordinator, a PITR recovery request and reading a log record of the MPP database. The method also includes determining a type of the log record and updating a transaction table when the type of the log record is an abort transaction or a commit transaction.

Growing community of inventors

Cupertino, CA, United States of America

Le Cai