Mailing-List: contact giraph-user-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: giraph-user@incubator.apache.org
Received-SPF: pass (athena.apache.org: domain of claudio.martella@gmail.com
 designates 209.85.210.175 as permitted sender)
MIME-Version: 1.0
In-Reply-To: <20111220152906.236690@gmx.net>
References: <20111220152906.236690@gmx.net>
From: Claudio Martella <claudio.martella@gmail.com>
Date: Tue, 20 Dec 2011 16:52:56 +0100
Message-ID: 
 <CAFJOoJfz+ehRAYtNdXTSzU4Xh4tEbM1OXEL6Nc4EOMmd+UtX3Q@mail.gmail.com>
Subject: Re: Is there a global state I can use?
To: giraph-user@incubator.apache.org
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Hi,

a general way of collecting data from all the vertices is using an
Aggregator. An aggregator collects messages from all the vertices (who
decide to write to it) and it can be read by all the vertices. You
could easily implement your statistics from there. Aggregators are
computed both on the workers and on the master, so it could be quite
scalable.

Hope it helps,
Claudio

On Tue, Dec 20, 2011 at 4:29 PM,  <listenbruder@gmx.net> wrote:
>
> Hi all,
>
> a plan to use Giraph for a use case where nodes send messages depending o=
n some global distribution of a value. For instance, nodes have a numeric v=
alue. Thus there is a global distribution of that value. Now I want all nod=
es to take an action, i.e., send messages, that have a value in say the top=
 1% of all values.
> How could I do this?
> Thinking in Hadoop MapReduce I'd use the distributed cache in order to ma=
intain a fingerprint of the global distribution.
> Would this work in giraph too?
>
> Thanks and BR!
> christoph


--=20
=A0 =A0Claudio Martella
=A0 =A0claudio.martella@gmail.com