Use bounded labels

A metric label should be a dimension you aggregate or filter by, with a bounded set of values.

The single most important rule is that label values must be bounded. A label like env has a handful of values forever. A label like user_id is an unbounded label: it grows without limit, and every new value creates another active series that’s billed and indexed.

Never use these as metric label values:

User, customer, or session IDs
Request or trace IDs
IP addresses
Email addresses
Timestamps

Each of these creates a new series per unique value, and the combinations multiply against every other label on the metric.

Read script

Script

The most important rule for metric labels is that label values must be bounded; they should have a limited number of unique values.

A good metric label is a dimension you actually aggregate or filter by, with a value set that stays small. An environment label has a handful of values, and it’ll still have a handful of values a year from now.

Compare that with a user ID label. That’s an unbounded label. It grows without limit, and every new value creates another active series that’s billed and indexed.

That’s why our never-use list includes user, customer, and session IDs, request and trace IDs, IP addresses, email addresses, and timestamps. Each of these creates a new series per unique value, and those combinations multiply against every other label on the metric.

Use bounded labels

Still have questions?

Get every update