You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@reef.apache.org by John Yang <jo...@gmail.com> on 2016/04/01 00:21:56 UTC

Re: [DISCUSS] Apache REEF is ...

Hi Markus,

I like Option E. It's better to use as few words as possible and emphasize
its role as a standard library.

Thanks,
John
On Apr 1, 2016 6:58 AM, "Markus Weimer" <ma...@weimo.de> wrote:

> Hi,
>
> we frequently have the need to summarize REEF in a single sentence, e.g.
> in our README.md, talk announcements, board reports etc. We have
> iterated through many ways of doing that in the past. However, I think
> it might help to drive some consensus around this.
>
> Our current README.md proclaims (Option A):
>
> Apache REEF™ (Retainable Evaluator Execution Framework) is a
>> scale-out computing fabric that simplifies the development of Big
>> Data applications on top of resource managers (e.g., Apache YARN and
>> Mesos).
>>
>
> I don't think that is terribly accurate. Sergey suggested as part of
> REEF-1252 to change it to something like (Option B):
>
> Apache REEF is a standard library for building distributed systems
>> that simplifies the development of Big Data applications on top of
>> resource managers (e.g., Apache Hadoop YARN and Apache Mesos).
>>
>
> Our website suggests (Option C):
>
> REEF, the Retainable Evaluator Execution Framework, is our approach
>> to simplify and unify the lower layers of big data systems on modern
>> resource managers.
>>
>
>
> And the SIGMOD paper uses the following text in its abstract (Option D):
>
> This paper presents REEF, a development framework that provides a
>> control-plane for scheduling and coordinating task-level (data-plane)
>> work on cluster resources obtained from a Resource Manager.
>>
>
> Of those, I like option B the best, but would propose a slight
> simplification (Option E):
>
> Apache REEF is a standard library for building distributed
>> applications on top of cluster resource managers (e.g., Apache Hadoop
>> YARN and Apache Mesos).
>>
>
>
> What do you think?
>
> Markus
>

Re: [DISCUSS] Apache REEF is ...

Posted by Dongjoon Hyun <do...@apache.org>.
Oh, what I meant was that someday we don't need any Cluster Resource
Manager at all.

For developers, REEF seems some kind of agnostic library in terms of
resource managers.

However, never mind this. We didn't have Standalone-runtime so far. :)

Dongjoon.

On Fri, Apr 1, 2016 at 11:23 AM, Markus Weimer <ma...@weimo.de> wrote:

> On 3/31/2016 3:32 PM, Dongjoon Hyun wrote:
>
>> By the way, we are working on `Standalone Runtime`. [...]
>> I'm not sure, but it seems not to include that fully. Do you think it is
>> good we expose that concept here, too?
>>
>
> Hmm, I haven't thought about it that way. When mentioning YARN and Mesos,
> I was going for name recognition, mostly. I believe explaining REEF's
> portability to other resource managers, including the standalone runtime,
> can happen in the sentences that follow.
>
> Makes sense?
>
> Markus
>

Re: [DISCUSS] Apache REEF is ...

Posted by Markus Weimer <ma...@weimo.de>.
On 3/31/2016 3:32 PM, Dongjoon Hyun wrote:
> By the way, we are working on `Standalone Runtime`. [...]
> I'm not sure, but it seems not to include that fully. Do you think it is
> good we expose that concept here, too?

Hmm, I haven't thought about it that way. When mentioning YARN and 
Mesos, I was going for name recognition, mostly. I believe explaining 
REEF's portability to other resource managers, including the standalone 
runtime, can happen in the sentences that follow.

Makes sense?

Markus

Re: [DISCUSS] Apache REEF is ...

Posted by Dongjoon Hyun <do...@apache.org>.
Hi, Markus.

+1 for Option E.

By the way, we are working on `Standalone Runtime`. (Hopefully, at REEF
0.15 ?)
I'm not sure, but it seems not to include that fully. Do you think it is
good we expose that concept here, too?

Dongjoon.


On Thu, Mar 31, 2016 at 3:21 PM, John Yang <jo...@gmail.com> wrote:

> Hi Markus,
>
> I like Option E. It's better to use as few words as possible and emphasize
> its role as a standard library.
>
> Thanks,
> John
> On Apr 1, 2016 6:58 AM, "Markus Weimer" <ma...@weimo.de> wrote:
>
> > Hi,
> >
> > we frequently have the need to summarize REEF in a single sentence, e.g.
> > in our README.md, talk announcements, board reports etc. We have
> > iterated through many ways of doing that in the past. However, I think
> > it might help to drive some consensus around this.
> >
> > Our current README.md proclaims (Option A):
> >
> > Apache REEF™ (Retainable Evaluator Execution Framework) is a
> >> scale-out computing fabric that simplifies the development of Big
> >> Data applications on top of resource managers (e.g., Apache YARN and
> >> Mesos).
> >>
> >
> > I don't think that is terribly accurate. Sergey suggested as part of
> > REEF-1252 to change it to something like (Option B):
> >
> > Apache REEF is a standard library for building distributed systems
> >> that simplifies the development of Big Data applications on top of
> >> resource managers (e.g., Apache Hadoop YARN and Apache Mesos).
> >>
> >
> > Our website suggests (Option C):
> >
> > REEF, the Retainable Evaluator Execution Framework, is our approach
> >> to simplify and unify the lower layers of big data systems on modern
> >> resource managers.
> >>
> >
> >
> > And the SIGMOD paper uses the following text in its abstract (Option D):
> >
> > This paper presents REEF, a development framework that provides a
> >> control-plane for scheduling and coordinating task-level (data-plane)
> >> work on cluster resources obtained from a Resource Manager.
> >>
> >
> > Of those, I like option B the best, but would propose a slight
> > simplification (Option E):
> >
> > Apache REEF is a standard library for building distributed
> >> applications on top of cluster resource managers (e.g., Apache Hadoop
> >> YARN and Apache Mesos).
> >>
> >
> >
> > What do you think?
> >
> > Markus
> >
>