Cardinality Counting in Redis

Cardinality counting is used to calculate the amount of elements without any duplication. In Redis, there are many data structures able to accomplish this job. However, what is the most applicable way for your use cases? This article will show the consideration behind the technical selection.

User Scenario

Suppose we need to get the failure rate in a sensor network to investigate the reporting qualities. Therefore, we have to record the health status in hours by the incoming requests.

The key point is to simplify the process, we don’t want to get the value first, determine whether it is existing, and then insert the record like:

Instead, we should insert the record every time, and the storage can de-duplicate for us. Or, we can limited pre-process data to make the storage do faster.

Assume that we have a sensor A, and the sensor requested to the server at 1/2 1:11, 1/3 2:22, and 1/8 3:00.

Alright, let’s see how Redis did cardinality counting.

Set

The basic idea is using set. Before adding to the set, we have to pre-process the date. Due to our requirement, we only keep the hour without minutes and seconds.

const date1 = new Date(2021, 0, 2, 1, 0);
const d1 = date1.toISOString();