公开(公告)号：US11748155B1

公开(公告)日：2023-09-05

申请号：US17659925

申请日：2022-04-20

Applicant: Snowflake Inc.

Inventor： Yuchen Cao ,  Stephen Mark Andrew Clark ,  Supriya Vasudevan ,  Jinzhou Yang

IPC: G06F9/48 ,  G06F8/41 ,  G06F8/30

CPC classification number: G06F9/4881 ,  G06F8/41 ,  G06F8/311

Abstract: Provided herein are systems and methods for processing a script file generated using a declarative engine for workloads (DEW). A system includes at least one hardware processor coupled to a memory and configured to retrieve a script file. The script file specifies a plurality of tasks forming a plan and an order of execution for the plurality of tasks. The script file is compiled to generate compiled code. The compiled code is executed to generate an in-memory representation of the plan. The in-memory representation of the plan is converted into a graph such as a directed acyclic graph (DAG). The graph includes the plurality of tasks. The plurality of tasks in the graph are executed according to the order of execution.

公开(公告)号：US11467945B1

公开(公告)日：2022-10-11

申请号：US17654884

申请日：2022-03-15

Applicant: Snowflake Inc.

Inventor： Yuchen Cao ,  Stephen Mark Andrew Clark ,  Prasanna Kumar Krishnamurthy ,  Supriya Vasudevan ,  Jinzhou Yang

IPC: G06F11/00 ,  G06F11/36 ,  G06F16/21

Abstract: Provided herein are systems and methods for resilience testing. A system includes at least one hardware processor coupled to a memory and configured to decode a workflow to obtain a workload specification and a failure experiment specification. A first set of containers is configured to execute one or more workloads on a testing node. The one or more workloads are defined by the workload specification. A second set of containers is configured to execute one or more failure experiments on the testing node. The one or more failure experiments are based on the failure experiment specification. Execution of the one or more failure experiments triggers an error condition on the testing node. A notification is generated based on at least one metric associated with execution of the one or more workloads and the one or more failure experiments.

公开(公告)号：US11762702B1

公开(公告)日：2023-09-19

申请号：US17660055

申请日：2022-04-21

Applicant: Snowflake Inc.

Inventor： Yuchen Cao ,  Stephen Mark Andrew Clark ,  Supriya Vasudevan ,  Jinzhou Yang

IPC: G06F9/50 ,  G06F9/48

CPC classification number: G06F9/5038 ,  G06F9/4881

Abstract: Provided herein are systems and methods for resilience testing using a DEW. A system includes at least one hardware processor coupled to a memory and configured to decode a workflow to obtain a plurality of workflow steps. Each workflow step of the plurality of workflow steps includes a runtime engine identifier and a script identifier. A script for a workflow step of the plurality of workflow steps is retrieved based on the script identifier. The script specifies a plurality of tasks. One or more containers are configured with a runtime engine based on the runtime engine identifier. The script executes against the one or more containers using the runtime engine, to perform the plurality of tasks.

公开(公告)号：US11409632B1

公开(公告)日：2022-08-09

申请号：US17652902

申请日：2022-02-28

Applicant: Snowflake Inc.

Inventor： Yuchen Cao ,  Stephen Mark Andrew Clark ,  Prasanna Kumar Krishnamurthy ,  Supriya Vasudevan ,  Jinzhou Yang

IPC: G06F11/00 ,  G06F11/36 ,  G06F16/21

Abstract: Provided herein are systems and methods for resilience testing. A system includes at least one hardware processor coupled to a memory and configured to decode a workflow to obtain a workload specification and a network experiment specification. A first set of containers is configured to execute one or more workloads on a testing node. The one or more workloads are defined by the workload specification. A second set of containers is configured to execute one or more network experiments on the testing node. The one or more network experiments are based on the network experiment specification. Execution of the one or more network experiments triggers an error condition on the testing node. A notification is generated based on at least one metric associated with execution of the one or more workloads and the one or more network experiments.