Mailing-List: contact yarn-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: yarn-issues@hadoop.apache.org
Date: Thu, 18 Jun 2015 19:38:01 +0000 (UTC)
From: "Wei Shao (JIRA)" <jira@apache.org>
To: yarn-issues@hadoop.apache.org
Message-ID: <JIRA.12838037.1434426977000.112611.1434656281445@Atlassian.JIRA>
In-Reply-To: <JIRA.12838037.1434426977000@Atlassian.JIRA>
References: <JIRA.12838037.1434426977000@Atlassian.JIRA>
 <JIRA.12838037.1434426977903@arcas>
Subject: [jira] [Commented] (YARN-3806) Proposal of Generic Scheduling
 Framework for YARN
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable


    [ https://issues.apache.org/jira/browse/YARN-3806?page=3Dcom.atlassian.=
jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=3D14592=
384#comment-14592384 ]=20

Wei Shao commented on YARN-3806:
--------------------------------

Hi Karthik, see proposal and comments above for why this proposal is came u=
p. I am aware of FairScheduler design and proposal of YARN-3306. Thanks!

> Proposal of Generic Scheduling Framework for YARN
> -------------------------------------------------
>
>                 Key: YARN-3806
>                 URL: https://issues.apache.org/jira/browse/YARN-3806
>             Project: Hadoop YARN
>          Issue Type: Improvement
>          Components: scheduler
>            Reporter: Wei Shao
>         Attachments: ProposalOfGenericSchedulingFrameworkForYARN-V1.2.pdf
>
>
> Currently, a typical YARN cluster runs many different kinds of applicatio=
ns: production applications, ad hoc user applications, long running service=
s and so on. Different YARN scheduling policies may be suitable for differe=
nt applications. For example, capacity scheduling can manage production app=
lications well since application can get guaranteed resource share, fair sc=
heduling can manage ad hoc user applications well since it can enforce fair=
ness among users. However, current YARN scheduling framework doesn=E2=80=99=
t have a mechanism for multiple scheduling policies work hierarchically in =
one cluster.
> YARN-3306 talked about many issues of today=E2=80=99s YARN scheduling fra=
mework, and proposed a per-queue policy driven framework. In detail, it sup=
ported different scheduling policies for leaf queues. However, support of d=
ifferent scheduling policies for upper level queues is not seriously consid=
ered yet.=20
> A generic scheduling framework is proposed here to address these limitati=
ons. It supports different policies (fair, capacity, fifo and so on) for an=
y queue consistently. The proposal tries to solve many other issues in curr=
ent YARN scheduling framework as well.
> Two new proposed scheduling policies YARN-3807 & YARN-3808 are based on g=
eneric scheduling framework brought up in this proposal.


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