Can Hive be configured to work with multiple namenodes(clusters)? I
understand we can use command 'SET' to set any hadoop (or hive)
configuration variable. But is it possible to handle multiple cluster
together ?

I'm trying to figure out, if its possible to join tables from different
Hadoop clusters.

--Dani
Sent from my &roid, pls forgive brevity.

Hive does not 'join' your data, it's all done by hadoop.

On Sat, Mar 17, 2012 at 7:27 AM, Dani Rayan <[EMAIL PROTECTED]> wrote:
> Can Hive be configured to work with multiple namenodes(clusters)? I
> understand we can use command 'SET' to set any hadoop (or hive)
> configuration variable. But is it possible to handle multiple cluster
> together ?
>
> I'm trying to figure out, if its possible to join tables from different
> Hadoop clusters.
>
> --Dani
> Sent from my &roid, pls forgive brevity.

I understand Hive submits translated MR to "a" jobtracker. My end goal is
generic, to re-iterate : "I'm trying to figure out, if its possible to join
tables from different Hadoop clusters." . (Without moving data) Using
something or trying to write own wrapper ?

On Sat, Mar 17, 2012 at 6:31 AM, wd <[EMAIL PROTECTED]> wrote:

> Hive does not 'join' your data, it's all done by hadoop.
>
> On Sat, Mar 17, 2012 at 7:27 AM, Dani Rayan <[EMAIL PROTECTED]> wrote:
> > Can Hive be configured to work with multiple namenodes(clusters)? I
> > understand we can use command 'SET' to set any hadoop (or hive)
> > configuration variable. But is it possible to handle multiple cluster
> > together ?
> >
> > I'm trying to figure out, if its possible to join tables from different
> > Hadoop clusters.
> >
> > --Dani
> > Sent from my &roid, pls forgive brevity.
>

--
-Dani Abel Rayan

Hi Dani
      When you say that your data is on two hadoop clusters, it means a Mapreduce job has to spawn its tasks across 2 or more clusters. AFAIK atm this is out of scope of MR frame work. So the answer is no. You can't materialize such a join with hive.

Regards
Bejoy.K.S
________________________________
 From: Dani Rayan <[EMAIL PROTECTED]>
To: [EMAIL PROTECTED]
Sent: Saturday, March 17, 2012 9:55 PM
Subject: Re: Hive working with multiple clusters
 
I understand Hive submits translated MR to "a" jobtracker. My end goal is generic, to re-iterate : "I'm trying to figure out, if its possible to join tables from different Hadoop clusters." . (Without moving data) Using something or trying to write own wrapper ?
On Sat, Mar 17, 2012 at 6:31 AM, wd <[EMAIL PROTECTED]> wrote:

Hive does not 'join' your data, it's all done by hadoop.
>
>
>On Sat, Mar 17, 2012 at 7:27 AM, Dani Rayan <[EMAIL PROTECTED]> wrote:
>> Can Hive be configured to work with multiple namenodes(clusters)? I
>> understand we can use command 'SET' to set any hadoop (or hive)
>> configuration variable. But is it possible to handle multiple cluster
>> together ?
>>
>> I'm trying to figure out, if its possible to join tables from different
>> Hadoop clusters.
>>
>> --Dani
>> Sent from my &roid, pls forgive brevity.
>
--
-Dani Abel Rayan