Data delay detection method and apparatus, electronic device, and computer-readable medium

Patent metadata

Field	Value
Publication number	US-12436797-B2
Application number	US-202418967974-A
Country	US
Kind code	B2
Filing date	Dec 4, 2024
Priority date	Dec 19, 2023
Publication date	Oct 7, 2025
Grant date	Oct 7, 2025

Abstract

Provided are a data delay detection method and apparatus, an electronic device, and a computer-readable medium. The method includes: first, obtaining a baseline construction parameter; then, constructing a baseline instance according to the baseline construction parameter; then, obtaining a data delay detection result of each real-time task; and finally, determining delay state characterization data of the baseline instance according to the data delay detection results of the at least two real-time tasks, so that the delay state characterization data can better represent a data delay of a full-link task used to process the real-time data stream.

First claim

What is claimed:

1. A data delay detection method, comprising:
obtaining a baseline construction parameter;
constructing a baseline instance according to the baseline construction parameter, wherein the baseline instance comprises a task link graph and a node parameter of each node in the task link graph, the task link graph is used to describe a data dependency relationship between at least two real-time tasks, different nodes in the task link graph are used to represent different real-time tasks, and the at least two real-time tasks are used to process a real-time data stream;
obtaining a data delay detection result of each real-time task; and
determining delay state characterization data of the baseline instance according to data delay detection results of the at least two real-time tasks, wherein
the baseline construction parameter comprises a task identifier of a baseline guarantee task, a baseline commitment delay, a baseline early warning delay, and a task delay gain coefficient, and
the constructing a baseline instance according to the baseline construction parameter comprises:
determining the at least two real-time tasks and the data dependency relationship between the at least two real-time tasks according to a pre-constructed task link relationship and the task identifier of the baseline guarantee task, wherein the task link relationship comprises the data dependency relationship between the at least two real-time tasks, and the at least two real-time tasks comprise the baseline guarantee task and a task that directly or indirectly provides data for the baseline guarantee task; and
constructing the baseline instance according to the at least two real-time tasks, the data dependency relationship between the at least two real-time tasks, the baseline commitment delay, the baseline early warning delay, and the task delay gain coefficient.

2. The method according to claim 1 , wherein the constructing the baseline instance comprises:
constructing, in the task link graph, a node corresponding to the baseline guarantee task, and determining a node parameter of the node corresponding to the baseline guarantee task according to the baseline commitment delay and the baseline early warning delay;
determining, according to the data dependency relationship between the at least two real-time tasks, an upstream task of a target task from the at least two real-time tasks, wherein an initial value of the target task is the baseline guarantee task, and the data dependency relationship between the at least two real-time tasks comprises a data dependency relationship between the target task and the upstream task;
constructing, in the task link graph, a node corresponding to the upstream task, constructing, in the task link graph, a connection relationship between the node corresponding to the upstream task and the node corresponding to the target task, and determining a node parameter of the node corresponding to the upstream task according to the task delay gain coefficient and the node parameter of the node corresponding to the target task; and
updating a traversal description state of the node corresponding to the target task to a traversed state, updating the target task with a task corresponding to a node in an untraversed state in the task link graph, and continuing to perform the operation of determining, according to the data dependency relationship between the at least two real-time tasks, an upstream task of a target task from the at least two real-time tasks, until all nodes in the task link graph are in the traversed state.

3. The method according to claim 2 , wherein after the determining, from the at least two real-time tasks, an upstream task of a target task, the method further comprises:
determining whether there is a node corresponding to the upstream task in the task link graph; and
if there is the node corresponding to the upstream task in the task link graph, constructing, in the task link graph, a connection relationship between the node corresponding to the upstream task and the node corresponding to the target task, and updating a node parameter of the node corresponding to the upstream task according to the task delay gain coefficient and the node parameter of the node corresponding to the target task; and
the constructing, in the task link graph, a node corresponding to the upstream task comprises:
if there is no node corresponding to the upstream task in the task link graph, constructing, in the task link graph, the node corresponding to the upstream task.

4. The method according to claim 2 , wherein the node parameter comprises a task delay constraint,
a task delay constraint of the node corresponding to the baseline guarantee task comprises the baseline commitment delay and the baseline early warning delay, and
a task delay constraint of the node corresponding to the upstream task is determined according to a product of the task delay gain coefficient and a task delay constraint of the node corresponding to the target task.

5. The method according to claim 2 , wherein the node parameter comprises a task hierarchy;
a task hierarchy of the node corresponding to the baseline guarantee task is a preset hierarchy; and
a task hierarchy of the node corresponding to the upstream task is determined according to a sum of a preset hierarchy step and a task hierarchy of the node corresponding to the target task.

6. The method according to claim 2 , wherein the node parameter comprises a task hierarchy;
nodes in an untraversed state in the task link graph comprise at least one candidate node; and
the updating the target task with a task corresponding to a node in an untraversed state in the task link graph comprises:
selecting a to-be-used node from the at least one candidate node according to a task hierarchy of each candidate node, wherein a task hierarchy of the to-be-used node is not higher than a task hierarchy of another node other than the to-be-used node in the at least one candidate node; and
updating the target task with the task corresponding to the to-be-used node.

7. The method according to claim 2 , wherein the node parameter of the node corresponding to the upstream task comprises an impact task identifier;
if the target task is the baseline guarantee task, the impact task identifier of the node corresponding to the upstream task is determined according to the task identifier of the target task; or
if the target task is not the baseline guarantee task, the impact task identifier of the node corresponding to the upstream task is determined according to the impact task identifier of the node corresponding to the target task.

8. The method according to claim 1 , wherein the node parameter comprises a task delay constraint, and
the determining delay state characterization data of the baseline instance according to the data delay detection results of the at least two real-time tasks comprises:
determining the delay state characterization data of the baseline instance according to the data delay detection results of the at least two real-time tasks and the task delay constraint of each node in the task link graph.

9. The method according to claim 8 , wherein the at least two real-time tasks comprise at least one baseline guarantee task,
the delay state characterization data comprises a baseline state, and
the determining a baseline state comprises:
for any baseline guarantee task, determining a data delay state corresponding to the baseline guarantee task according to a comparison result between a data delay detection result of the baseline guarantee task and a task delay constraint of a node used to represent the baseline guarantee task; and
selecting a target state from data delay

Assignees

• Beijing Volcano Engine Technology Co Ltd

Inventors

• Meng He
• Zhang Huarui
• Min Jia
• Li Zhou

Classifications

• G06F9/54

  Interprogram communication · CPC title

• G06F9/466 Primary

  Transaction processing · CPC title

• G06F9/5038 Primary

  considering the execution order of a plurality of tasks, e.g. taking priority or time dependency constraints into consideration (scheduling strategies G06F9/4881 and subgroups) · CPC title