TensorFlow Serving MonitoringConfig: Configure and Monitor Your Model Serving Setup

MonitoringConfig is a configuration object in TensorFlow Serving that allows users to specify various monitoring options for their serving setup. It provides options for configuring the monitoring of different metrics and setting up monitoring exporters.\n\nSome of the properties and methods available in MonitoringConfig are:\n\n- collection_period: The time interval at which monitoring metrics are collected. It is specified in seconds.\n- exporters: A list of monitoring exporters to be used. Each exporter is a separate configuration object that defines the details of the exporter, such as its type, address, and credentials.\n- metric_config: A configuration object that specifies the metrics to be monitored. It allows users to define the names and types of the metrics they want to monitor, along with any additional properties specific to each metric.\n- monitoring_level: An enum that defines the level of monitoring to be performed. It can be set to MINIMAL, LOW, MEDIUM, or HIGH, with each level providing different levels of monitoring granularity.\n\nUsers can create an instance of MonitoringConfig and set its properties according to their monitoring requirements. This configuration object is then passed to the TensorFlow Serving server during startup to enable monitoring for the serving setup.