We need to to deploy InfluxDB for anomaly detection in streaming data using Influx, Telegraf, and Kapacitor on K8s.
Are we required to deploy each one as separate installations or can the use new version of Influx OSS handle all of that? If possible, could you provide some sample YAML for the deployment?
Also, According to documentation Kubernetes | InfluxData Platform Documentation
Influx in kubernetes is not recommended. Can we use it standalone for the above.
Hello @Utkarsh_Kumar,
Welcome!
I wouldn’t recommend using Kapacitor. It doesn’t have support.
I’d hold off on using InfluxDB v3 until the OSS and Community editions come out.
Migrating from v1 to v3 will be easier.
Arguments in favor of using v3:
InfluxDB 3.0 is up to 45x Faster for Recent Data Compared to InfluxDB Open Source | InfluxData.
If you’re interested in deploying InfluxDB with K8s I suggest looking into:
For anomaly detection using InfluxDB v3 I would aslo recommend the following resources:
Using QUIX:
Using Mage.ai:
You can use the same tool and idea but add your anomaly detection instead.
Use the ADTK tool kit python package with MQTT and Telegraf and Grafana:
For that example navigate to localhost:5005 to turn anomalies on and off.
A simple example of using Kafka and InfluxDB and Telegraf:
Kafka, MQTT, and InfluxDB:
These are all pretty simple examples but hopefully they’ll get you started with the general architecture idea for building an anomaly detection pipeline.
Basically regardless of whether you’re using v1 or v3 you’ll want to use some other tools with a client library ( I recommend python) for the actual anomaly detection
