[ 
https://issues.apache.org/jira/browse/AURORA-1384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14621265#comment-14621265
 ] 

Stephan Erb commented on AURORA-1384:
-------------------------------------

There is an interesting talk from the Netflix guys from last year's Mesoscon. 
They are describing a more general scheduling system consisting of hard- and 
soft-constraints. The binpacking quality you desire is such a soft-constraint.

If I remember correctly, the idea boils down to this:

* Select a subset of slaves (as doing it for all of them might be expensive)
* Compute the weighted fitness of a task in regard to a slave, i.e., how good 
does fit according to various criteria such as binpacking. Binpacking is as 
simple as `usedCPUs/totalCPUs`. You get the total fitness as 
`CpuBinPack*CPUWeight + MemBinPack*MemWeight + ...`
* Schedule the task on the slave with the highest fitness (given that this 
suits your threshold).

I'd prefer such a system over some hardcoded assignment policies, as it would 
also enabled uses cases such as the ones implied by AURORA-173.

> Provide alternate policies for offer assignment to tasks
> --------------------------------------------------------
>
>                 Key: AURORA-1384
>                 URL: https://issues.apache.org/jira/browse/AURORA-1384
>             Project: Aurora
>          Issue Type: Story
>          Components: Scheduler
>    Affects Versions: 0.8.0
>            Reporter: Anindya Sinha
>            Assignee: Anindya Sinha
>
> Currently, offers are arranged in a list. When new tasks need to be assigned 
> an offer, the first offer from the list that matches the task constraints is 
> assigned which over time may result in a non even distribution of tasks on 
> slaves in the cluster.
> In addition to existing algorithm ("first-fit"), we can support alternate 
> algorithms as follows:
> i) bin-packed: Offer is assigned to a task based on a bin packed algorithm 
> which ensures a specific offer is used up to the maximum possible with 
> respect to resources (cpu) before moving to an alternate offer.
> ii) even-distribution: Offer is assigned to ensure even distribution of load 
> on each slave on the cluster based on resources (cpu).



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

Reply via email to