Thanks for the article with interesting technical details!

It is interesting to know the following additional details:

• How many nodes does each InfluxDB cluster have and what is the hardware configuration of these nodes?
• How many unique time series do regularly receive new data points? I.e. what is the cardinality of the time series?

Also how do you deal with the extended cluster unavailability? Kafka queues for the unavailable cluster can quickly overflow the available storage at the incoming rate of 2M messages per second.