search

elastic / kibana

Your window into the Elastic Stack
https://www.elastic.co/products/kibana
Other
19.71k stars 8.13k forks source link

[Infra] Anomaly detection setup UI changes #189701

Open crespocarlos opened 1 month ago

crespocarlos commented 1 month ago

Summary

The anomaly detection set up UI needs to be updated. Note that the advanced settings fields will now be inside an accordion, which will be collapsed by default.

Copy update

Enable machine learning for (hosts|kubernetes)

Enable anomaly detection using our recommended settings or fine-tune your jobs using the advanced settings. If you need to make changes after creating your jobs, you can recreate them at any time (previous anomalies will be removed).

Copy update

1st field

When do your jobs begin? By default, the last 4 weeks of data will be used to build the anomaly detection model. Increasing the look-back will give the model longer to learn the periodicity in your data, but it will take longer for the job to analyze.

2nd field - HOSTS ONLY

How do you want to partition the analysis? You can optionally partition the data to build independent baselines for groups of data that share the same characteristics. For example, you might choose to partition by machine type or availability zone. Learn more (link: https://ela.st/infra-anomaly-partition)

3rd field - HOSTS ONLY

Which Hosts would you like to detect anomalies on? If you are using a partition, we would recommend applying a filtering to only evaluate the hosts that are really important to you to optimise performance. For example, you may apply a filter to only detect anomalies on a particular machine type so other machine types are not evaluated. Learn more (link: https://ela.st/infra-host-ad-filtering)

When Partition field is selected, we need to show the following badge next to the title

AC

elasticmachine commented 1 month ago

Pinging @elastic/obs-ux-infra_services-team (Team:obs-ux-infra_services)

jennypavlova commented 3 weeks ago

I will keep the draft PR as I already started working on that and it's almost ready so when we decide to prioritize the work on the issue I can continue and open the PR.

CC: @crespocarlos @roshan-elastic

smith commented 3 weeks ago

We're not prioritizing this at this time.

roshan-elastic commented 3 weeks ago

Hey @smith - do you think this is small enough to be a maintenance backlog candidate?

smith commented 3 weeks ago

@roshan-elastic that's not a problem. You can reopen, unarchive and move to backlog.