Post Metrics

Send data samples to Anodot using Metric 2.0/3.0 protocols. Both the version 2.0 and 3.0 protocols work with a schema - this means that the data is structured into key-value pairs and enables Anodot to better 'understand' and interact the metrics sent to it. The main difference between 2.0 and 3.0 is that 3.0 enable sending a 'watermark' notification to Anodot, enabling a better detection and assignment of data points to a time frame.

Use Anodot's sample code for a quick start

You can use Anodot's sample Python script to send metrics in Metrics 2.0 or 3.0 protocols. Refer to Anodot on GitHub for more details.

High Level Flow

1. Creating a Schema
2. Sending samples matching the schema
3. Sending watermarks timestamp to direct Anodot to process data samples up to the timestamp

FAQs

• Is the schema mandatory?

Yes. The schema defines the expected structure of metrics sent using the Metric 3.0 protocol. Defining a schema enables advanced capabilities such as deeper analysis and investigation, helping improve root cause analysis for anomalies.

• What happens if I send data in an invalid structure?

A post metrics request consists of an array containing up to 10,000 data samples. Some entries may be invalid and rejected while others are successfully processed. Review the response payload to identify faulty entries and the associated errors.

• What does the watermark do?

The initial watermark indicates that historical (retroactive) data ingestion is complete. Subsequent watermarks notify the platform that an additional chunk of data is ready for processing and that no more data is expected for that time span. This allows composite metrics and alerts to be calculated immediately.

• Is using the watermark mandatory?

While not mandatory, using watermarks is highly recommended. Although Anodot can process data using its built-in flushing mechanism, watermarks provide precise processing boundaries, resulting in more accurate anomaly detection and alerting.