Baidu IntelliEdge - Cloud Management Suite

The cloud management suite is composed of Node Management, Application Deployment, Configuration Management and Node Pre-configuration.

• Node Management

  • Monitor the edge nodes, including edge node status and node resource utilization (CPU and memory utilization).
  • Monitor the edge applications deployed on edge nodes, including application status and application resource utilization (CPU and memory utilization).

• Application Deployment

  • Manage the service arrangement of the application. An application supports one or more services. If it is the kube mode, a service corresponds to a container. If it is the process mode, a service corresponds to a binary program.
  • Manage the application deployment. It supports the directional deployment of applications through node tag matching, and the nodes meeting the deployment tag will be deployed to the edge application automatically.

• Configuration management

  • Manage the custom configuration. It supports online editing, including KV, ciphertext, and image registry access credential.
  • It supports importing the configuration from existing cloud services. For example, import the function package from Function Computing CFC and import the model file from AI Middle Platform BML and End Model Generation Platform EasyEdge.

• Node Pre-configuration

  • It supports the batch registration and activation of edge nodes.

Baidu IntelliEdge - Edge Local Run Package

Edge Local Run Package is mainly composed of two parts as follows:

• BIE Edge Computing Framework: It is provided by the open-source edge computing framework baetyl to carry out the basic edge management and edge application arrangement.

• BIE Edge Application: For the edge application based on the baetyl framework, the baetyl framework provides the Edge Application Development SDK. BIE provides the common edge applications, mainly including the following capabilities:

  • Function computing, which provides the function runtime to support the edge function computing and support the local operation of python27, python36, and node85 functions.
  • Stream computing, which provides the stream computing runtime and supports the local stream computing.

  • AI inference

    • It can realize the edge AI inference of a specific model through the video-infer module and function computing.
    • It can realize the easyedge-inference transformation model through the easyedge-inference module.
  • Data cache, through the broker module, the offline caching of messages and continuous transmission after disconnection are realized.
  • Data storage, which realizes the edge temporal data storage through the tsdb module.
  • Data collection, which provides the southward data collection module. For example, the Modbus module collects data of associated devices connecting with the edge node.
  • Configuration synchronization, which realizes the real-time synchronization between the cloud configuration and the edge through the system module, and distributes the cloud configuration change to the edge node automatically.

  • Remote data transfer

    • Remote message transfer: It supports transferring the edge message to the remote MQTT Broker through the MQTT protocol.
    • Remote file transfer: It supports transferring the file to the remote object storage, for example, baidu bos, aws s3, and aws s3-compliant object storage service, such as ceph and minio.