You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@bigtop.apache.org by "jay vyas (JIRA)" <ji...@apache.org> on 2015/12/02 16:53:11 UTC

[jira] [Comment Edited] (BIGTOP-2161) [bigpetstore] transaction-queue Implment/Brainstorm ideas for global data rate.

    [ https://issues.apache.org/jira/browse/BIGTOP-2161?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15035998#comment-15035998 ] 

jay vyas edited comment on BIGTOP-2161 at 12/2/15 3:52 PM:
-----------------------------------------------------------

we would like to keep the dependency on state minimal.  So that means very little message passing.  Every message usually would convey info about the state of another actor... Sort of opening a can of worms in terms of features that might get abused.

Maybe just an algorithm which is stateless, but which attenuated over time according to clock time.  If we assume global time is correct, this mechanism would allow for coordinated load changes


was (Author: jayunit100):
we would like to keep the dependency on state minimal.  So that means very little message passing.
Maybe just an algorithm which is stateless, but which attenuated over time according to clock time.

> [bigpetstore] transaction-queue Implment/Brainstorm ideas for global data rate.
> -------------------------------------------------------------------------------
>
>                 Key: BIGTOP-2161
>                 URL: https://issues.apache.org/jira/browse/BIGTOP-2161
>             Project: Bigtop
>          Issue Type: Improvement
>          Components: blueprints
>    Affects Versions: backlog
>            Reporter: jay vyas
>             Fix For: backlog
>
>
> - We right now are able to generate n transactions a second per daemon.  
> - However, I'd like to implement a global data rate, so that generators slow down/speed up at diffferent times. 
> Some ideas on how to do this. 
> - Embed a "rate" function into the transaction generator options... such that it randomly speeds up/slows down over time.   downside: there is no real global data rate here.  its just a bunch of coordinated data generators.  
> - Send a new parameter, a REST endpoint, which can be scraped to get a hint of global data rate.  The generators can use that endpoint to calibrate how fast/slow they should be going.
> [~rnowling] [~evans_ye] [~sekikn]  im leaning towards #2.  Any thoughts?



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