Summary Of The Google Pregel

["Edward J. Yoon" <ed...@apache.org>]

Hello,

I just wrote the summary of the google pregel -
http://blog.udanax.org/2010/06/summary-of-google-pregel.html

Hope this helps you!
And, Hama BSP? We're doing a great job! :)

-- 
Best Regards, Edward J. Yoon
edwardyoon@apache.org
http://blog.udanax.org

Re: Summary Of The Google Pregel

["Edward J. Yoon" <ed...@apache.org>]

Hi,

Pregel paper introduces specialized graph computing infrastructure but
its computing engine part is basically based on BSP model and IMO, its
BSP style computing engine part can be also used for matrix algorithms
that often better fit a message passing model.

So, we are focusing on the implementation of the BSP engine first,
expecting it will be more general. You can check the current progress
here: http://wiki.apache.org/hama/Architecture

And as you said, I believe some high level API could be implemented
based on Hama BSP in the future. :)

Thanks.

On Tue, Jun 15, 2010 at 11:43 PM, Andrew Milkowski
<an...@gmail.com> wrote:
> thanks Edward! that was a great summary, in what sense do you think
> Pregel differs from Hama in its current stage of development but also
> in the road map going forward?
>
> Another question is if Pregel utilizes macro-language (high level API)
> couldn't exactly derive example from the paper, for me this is really
> fascinating part of the whole endeavour to be able to encapsulate
> graph/matrix operations at the higher level, many that deal with M/R
> applications can tell you that even simplest Page rank computations
> require multi-stage M/R jobs, not complex in themselves but requiring
> input from application developers, sometimes for research/discovery
> high level shortcut would be very beneficial (one example immediately
> comes to mind is Pig)
>
> thanks much again
>
> On Tue, Jun 15, 2010 at 2:58 AM, Edward J. Yoon <ed...@apache.org> wrote:
>> Hello,
>>
>> I just wrote the summary of the google pregel -
>> http://blog.udanax.org/2010/06/summary-of-google-pregel.html
>>
>> Hope this helps you!
>> And, Hama BSP? We're doing a great job! :)
>>
>> --
>> Best Regards, Edward J. Yoon
>> edwardyoon@apache.org
>> http://blog.udanax.org
>>
>



-- 
Best Regards, Edward J. Yoon
edwardyoon@apache.org
http://blog.udanax.org

Re: Summary Of The Google Pregel

[Mark Kerzner <ma...@gmail.com>]

I second the praise, thank you Edward for making complex simple

Mark

On Tue, Jun 15, 2010 at 9:43 AM, Andrew Milkowski <andrewmilkowski@gmail.com
> wrote:

> thanks Edward! that was a great summary, in what sense do you think
> Pregel differs from Hama in its current stage of development but also
> in the road map going forward?
>
> Another question is if Pregel utilizes macro-language (high level API)
> couldn't exactly derive example from the paper, for me this is really
> fascinating part of the whole endeavour to be able to encapsulate
> graph/matrix operations at the higher level, many that deal with M/R
> applications can tell you that even simplest Page rank computations
> require multi-stage M/R jobs, not complex in themselves but requiring
> input from application developers, sometimes for research/discovery
> high level shortcut would be very beneficial (one example immediately
> comes to mind is Pig)
>
> thanks much again
>
> On Tue, Jun 15, 2010 at 2:58 AM, Edward J. Yoon <ed...@apache.org>
> wrote:
> > Hello,
> >
> > I just wrote the summary of the google pregel -
> > http://blog.udanax.org/2010/06/summary-of-google-pregel.html
> >
> > Hope this helps you!
> > And, Hama BSP? We're doing a great job! :)
> >
> > --
> > Best Regards, Edward J. Yoon
> > edwardyoon@apache.org
> > http://blog.udanax.org
> >
>

Re: Summary Of The Google Pregel

[Andrew Milkowski <an...@gmail.com>]

thanks Edward! that was a great summary, in what sense do you think
Pregel differs from Hama in its current stage of development but also
in the road map going forward?

Another question is if Pregel utilizes macro-language (high level API)
couldn't exactly derive example from the paper, for me this is really
fascinating part of the whole endeavour to be able to encapsulate
graph/matrix operations at the higher level, many that deal with M/R
applications can tell you that even simplest Page rank computations
require multi-stage M/R jobs, not complex in themselves but requiring
input from application developers, sometimes for research/discovery
high level shortcut would be very beneficial (one example immediately
comes to mind is Pig)

thanks much again

On Tue, Jun 15, 2010 at 2:58 AM, Edward J. Yoon <ed...@apache.org> wrote:
> Hello,
>
> I just wrote the summary of the google pregel -
> http://blog.udanax.org/2010/06/summary-of-google-pregel.html
>
> Hope this helps you!
> And, Hama BSP? We're doing a great job! :)
>
> --
> Best Regards, Edward J. Yoon
> edwardyoon@apache.org
> http://blog.udanax.org
>