Herval Freire created ZEPPELIN-2711:
---------------------------------------
Summary: Metrics for service monitoring
Key: ZEPPELIN-2711
URL: https://issues.apache.org/jira/browse/ZEPPELIN-2711
Project: Zeppelin
Issue Type: New Feature
Reporter: Herval Freire
I'm putting this out mainly to start a discussion on what approach we want to
take here.
For starter, we have two options in terms of distribution:
- support JMX
- support some sort of custom metrics mechanism
Following the lines of projects such as Presto, we can expose these as JSON for
quick consumption
The second point is how we want to store those metrics:
- Do we want metrics per execution?
- Do we want quantiles/aggregated stats?
I'm not sure what's the usual volume of data on a Zeppelin notebook - for our
internal purposes, individual stats may be enough.
Finally, there's the question as to which stats libraries to use. I'm currently
between airlift (in case we go for aggregated + expose as json).
Thoughts? Is this something the project is accepting contributions on (or has a
roadmap for)?
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
