[jira] [Updated] (STORM-899) CPU/Memory/Network metrics collection

["Rick Kellogg (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/STORM-899?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Rick Kellogg updated STORM-899:
-------------------------------
    Component/s: storm-core

> CPU/Memory/Network metrics collection
> -------------------------------------
>
>                 Key: STORM-899
>                 URL: https://issues.apache.org/jira/browse/STORM-899
>             Project: Apache Storm
>          Issue Type: New Feature
>          Components: storm-core
>            Reporter: Robert Joseph Evans
>
> There are many use cases where it would be great to know how much memory/cpu/network a particular bolt/spout is using, and how much of that network is being used to send data to specific bolt instances downstream.
> This can be used to improve the scheduling of a topology, and also to provide automatic elasticity to a topology.
> Collecting these metrics is not too difficult, but being able to get them on a per bolt.spout basis is difficult.  Perhaps STORM-891 will make it simple for us to support this.
> These metrics should flow back to nimbus.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)