公开(公告)号：US20220035829A1

公开(公告)日：2022-02-03

申请号：US16943314

申请日：2020-07-30

Applicant: salesforce.com, inc.

Inventor： Zhidong Ke ,  Utsavi Benani ,  Kevin Terusaki ,  Yifeng Liu ,  Percy Mehta ,  Heng Zhang

IPC: G06F16/25 ,  G06F16/28 ,  G06F16/23 ,  G06F16/22 ,  G06F9/54 ,  G06F11/07

Abstract: Managing mutations in a data lake environment. A mutation request to cause write operations that modify data objects or structures within an environment for collecting unformatted raw data is received. The environment has at least a data table and a notification table. An entry is written to the data table with a streaming job configured to receive and process the mutation request. Entries to the data table specify at least records indicating changes to objects in the environment based on ingestion processing for the environment for collecting unformatted raw data and based on the mutation request. A corresponding entry is written to the notification table in response to a successful write attempt to the data table. The notification table entry has information about data table entries for a specified period. At least one data consumer is notified that the data table has been modified.

公开(公告)号：US20220012259A1

公开(公告)日：2022-01-13

申请号：US16922989

申请日：2020-07-07

Applicant: salesforce.com, inc.

Inventor： Zhidong Ke ,  Kevin Terusaki ,  Yifeng Liu ,  Utsavi Benani ,  Heng Zhang

IPC: G06F16/25 ,  G06F16/23

Abstract: Techniques and mechanisms for ingesting data through an atomic transaction are disclosed. Raw data is received from multiple disparate sources to be consumed in an environment that does not support atomic write operations to data consumers. The environment has at least a data table and a notification table. A write to an entry in the data table having an associated version is attempted. The data table entry corresponds to the data to be consumed. A write to a corresponding entry to the notification table is attempted in response to a successful write attempt to the data table. The notification table entry includes information about the corresponding data table entry. The version associated with the data table is modified in response to successful writes of both the data table entry and the notification table entry. At least one data consumer is notified that the data table version has been modified.