Find cardinality explosions in metrics
Prompt that finds label cardinality in metrics and proposes cuts plus a budget. Paste it into an AI tool of your choice, which then finds cardinality explosions in metrics.
- Topic
- Spreadsheets & data
- Type
- Prompt
- Level
- Intermediate
- Time
- about 15 min
Results depend on the tool, the model and what you enter.
Result preview
Example: written by our editors, not real AI output.
http_reqs{status,path,user_id} Series: 12M (+4M / week) Top: user_id (high), path (medium) Propose: drop user_id, bucket path Budget: max 2M series / job Alert: +20 % series in 24 h
Fill in
What you type stays in this browser window. We store and send nothing.
- Prometheus
- cloud monitoring
- APM
- own warehouse
- mixed
- storage grows
- queries slow
- cost spike
- prevention
- after a new label
- user_id
- url
- free text
- large enums
- unclear
Your prompt
You have edited the prompt. Changes in the fields above no longer overwrite it.
This is an instruction for an AI, not the result itself.
Where this prompt works
These AI tools are a good fit for this prompt:
How to use it
- Fill in the fields or tap a suggestion. Leave out names, addresses and other personal data.
- Copy the finished prompt into an AI tool of your choice.
- Paste metric export, label cardinalities and cost hints, without personal raw values.
Why does this work?
A label with request URL creates millions of series. Without a budget you notice it on the invoice.
Tips
- Ban high cardinality in the metric policy, not only in the alert.