[
https://issues.apache.org/jira/browse/IVY-469?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12723614#action_12723614
]
Jon Schneider commented on IVY-469:
-----------------------------------
Tell me if this is too simplistic of a viewpoint:
The irony here is that the ivy caching mechanism will probably generally be
used to store artifacts on a local or networked drive somewhere. Later
retrieval using parallelized artifact "downloading" (file copying) will
actually cause performance to suffer since the operation really is a file I/O
bound process and not a CPU bound process.
At most, artifact download parallelizing would be useful the first time a set
of dependencies is downloaded from the net, and how often does that really
occur for a repository as it matures?
I wish there was a magic bullet for file copying parallelization speed. But
then my particular use of Ivy is more the exception than the rule.
Input?
Jon
> Parallelize artifact downloading
> --------------------------------
>
> Key: IVY-469
> URL: https://issues.apache.org/jira/browse/IVY-469
> Project: Ivy
> Issue Type: Improvement
> Components: Core
> Reporter: Xavier Hanin
>
> For the moment artifact downloading during dependencies resolution happens
> sequentially, each artifact being downloaded one after the other.
> It would be nice to improve Ivy resolve engine to be able to parallelize
> downloading to optimize performance, especially when using a set of
> repositories with heterogeneous performance.
> We should see how load balancing feature of IVY-468 could be leveraged by
> this new downloading strategy.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
