Contents x
    CloudCore Management of Edgeflows
    • 22 Dec 2023
    • 1 Minute to read
    • Print
    • Dark
      Light
    Contents

    CloudCore Management of Edgeflows

    • Updated on 22 Dec 2023
    • 1 Minute to read
    • Print
    • Dark
      Light
    Article summary

    An EdgeFlow performs GPU-accelerated inference for on-premises data sources, e.g., network cameras located on a production line can directly communicate with an EdgeFlow for near real-time inference. The EdgeFlow can communicate with Cogniac cloud infrastructure for downloading models, uploading statistics, and selecting images for feedback.

    Every EdgeFlow object has the following fields.

    Name

    Example

    Description

    gateway_id
    string

    "Ajr2t45p"

    A unique ID used to identify the EdgeFlow.

    name
    string

    "Headquarters office gateway"

    EdgeFlow name, should be brief and descriptive.

    description
    string

    "Security cameras gateway"

    EdgeFlow description should be a brief yet complete description of the focus of this gateway.

    location
    string

    "cogniac-hq"

    Location of the EdgeFlow.

    poll_interval
    integer

    20

    The interval at which updates to the EdgeFlow configuration are implemented in the system in seconds.
     
     It defaults to 20 seconds.
     
     Accepts values greater than or equal to 1 second.

    model
    string

    EdgeFlow-RM-M20

    The EdgeFlow's device model.

    mac_address
    string

    "01-23-45-67-89-ab"

    The media access control (MAC) address of the gateway device. The MAC address is obtained from the first Ethernet device discovered on the EdgeFlow. The value is immutable.

    serial_number
    string

    "1234567-890-aabb"

    The serial number of the EdgeFlow device.

    ip_address
    string

    "123.45.67.89"

    The IP of the gateway device. Assigned based on the IP address of the requesting device during EdgeFlow creation.

    Model Deployment Policy
    string

    "latest"

    This field is used to control which models the EdgeFlow uses for inference. There are 3 options:
     "latest" - use current best model, "production" - use production models specified in app configuration, or "staging" - use staging models as specified in-app configuration.
     The default is "latest".

    created_at
    float

    1463179215.124683

    Unix timestamp.

    modified_at
    float

    1463179215.124683

    Unix timestamp.

    created_by
    string

    "admin-user@cognicac.co"

    The user that created the EdgeFlow.

    tenant_id
    string

    "rt06diepwc3i"

    The tenant ID for the EdgeFlow.

    input subjects
    string

    "input_subject_1"

    A list of input subjects that can be used to invoke /1/process/<input_subject_uid> calls to the local inference API.
     If the input subject is not specified, the /1/process/<input_subject_uid> api will return a 404 - subject not found error.

    There are additional fields when querying the status of a Gateway. They are listed on Gateways - Status page.

    Was this article helpful?
    What's Next