You are viewing a plain text version of this content. The canonical link for it is here.

Posted to mapreduce-user@hadoop.apache.org by Ann Pal <an...@yahoo.com> on 2011/12/20 19:56:48 UTC

resource allocation in Yarn:

Hi,
When resources are allocated in Map reduce Next gen, it can be based on cpu, memory, disk and network bandwidth.

Is network bandwidth the bandwidth from server to the switch (TOR) it is connected to?

Documentation says only memory is implemented. 
Are the other parameters planned as well. The Node Manager needs to implement these?

Thanks!

Re: resource allocation in Yarn:

Posted by Robert Evans <ev...@yahoo-inc.com>.

Ann,

Currently only memory is supported, like the documentation says.  There are plans to add in more, but none are actually in the code, and as such I am not sure if anyone can really comment on how this feature will work once/if it is implemented.  If you have some use case for it, and some direction you would like to see it go in, then please file a JIRA for it, and add in your comments to the JIRA.

--Bobby Evans

On 12/20/11 12:56 PM, "Ann Pal" <an...@yahoo.com> wrote:

Hi,
When resources are allocated in Map reduce Next gen, it can be based on cpu, memory, disk and network bandwidth.

Is network bandwidth the bandwidth from server to the switch (TOR) it is connected to?

Documentation says only memory is implemented.
Are the other parameters planned as well. The Node Manager needs to implement these?

Thanks!