Mailing-List: contact dev-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@cassandra.apache.org
Received-SPF: pass (athena.apache.org: local policy)
Content-Type: text/plain; charset=us-ascii
Mime-Version: 1.0 (Apple Message framework v1078)
Subject: Re: Hadoop package exposed through thrift
From: aaron morton <aaron@thelastpickle.com>
In-Reply-To: <894A9B59-AE42-4041-9552-DED01EA6CC39@gmail.com>
Date: Wed, 9 Jun 2010 23:22:35 +1200
Content-Transfer-Encoding: quoted-printable
Message-Id: <78E26A09-1FE6-4541-BBB1-C366FC8B8301@thelastpickle.com>
References: <29259B20-48D1-411A-87B7-0BA828D9124A@gmail.com>
 <AANLkTilkrE8JBtA3JpHgIRp9sdD3FCG1MLgOJymlk3sv@mail.gmail.com>
 <894A9B59-AE42-4041-9552-DED01EA6CC39@gmail.com>
To: dev@cassandra.apache.org

I'm not up to speed with Hadoop in Cassandra, but regular Hadoop =
provides a IO stream interface so it can be used with non Java =
languages.=20

http://hadoop.apache.org/common/docs/r0.15.2/streaming.html

That may be of help.=20
Aaron

On 9 Jun 2010, at 09:53, Jeremy Hanna wrote:

> I just didn't know if there were any way to make it easier for the =
non-java crowd to take advantage of it.  I'll give it some more thought.
>=20
> On Jun 8, 2010, at 4:05 PM, Jonathan Ellis wrote:
>=20
>> exposing it through thrift would mean the path would be
>>=20
>> client
>> to cassandra [processing thrift command]
>> to hadoop [giving it a job]
>> to cassandra [fetching the data]
>> to hadoop [m/r]
>> to cassandra [handing result back]
>> to client
>>=20
>> it just doesn't seem like a good design to me.
>>=20
>> additionally, thrift is meant more for "stuff your app is doing
>> constantly" while hadoop handles analytics queries.  this separation
>> of duties makes a lot of sense to me.
>>=20
>> On Tue, Jun 8, 2010 at 1:45 PM, Jeremy Hanna =
<jeremy.hanna1234@gmail.com> wrote:
>>> When I gave a presentation on cassandra+hadoop, some ruby folks were =
wondering about the possibility of using the MapReduce functionality in =
a language other than Java.
>>>=20
>>> I was just wondering if any thought was given to exposing the =
org.apache.cassandra.hadoop functionality through thrift.  That way the =
MapReduce code could be used by several languages and secondarily by =
client authors.
>>>=20
>>> I'm just trying to see if there is any reason why it wasn't exposed =
through thrift or if more needs to be done before it could be exposed to =
languages other than Java.
>>>=20
>>> Thanks,
>>>=20
>>> Jeremy
>>=20
>>=20
>>=20
>> --=20
>> Jonathan Ellis
>> Project Chair, Apache Cassandra
>> co-founder of Riptano, the source for professional Cassandra support
>> http://riptano.com
>=20