incubator-drill-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Clark Yang (杨卓荦) <yangzhuo...@gmail.com>
Subject Re: support for hadoop 1.X?
Date Mon, 17 Sep 2012 08:46:34 GMT
I think drill should support Hadoop and not limited on Hadoop.
Currenly, most basic data are maintained on the Hadoop cluster. Drill
should support load them and process.
BTW, Will drill has its own DFS or just HDFS?
If hdfs, I think there will be a great many limits on data locaily and
in-place computing.

Cheers,
Zhuoluo (Clark) Yang


2012/9/17 Ted Dunning <ted.dunning@gmail.com>:
> It is even conceivable that we could graft drill onto something like Storm
> to allow real-time munching of tuples.
>
> On Mon, Sep 17, 2012 at 1:16 AM, Azuryy Yu <azuryyyu@gmail.com> wrote:
>
>> yep, It should not limited on hadoop, Drill can be deployed as single mode
>> or distributed mode.
>>
>> but support hadoop/mapr  is necessary, but not limited.
>>
>>
>> On Mon, Sep 17, 2012 at 1:33 PM, Ted Dunning <ted.dunning@gmail.com>
>> wrote:
>>
>> > No.  I don't think that anybody has considered limiting Drill to Hadoop
>> > 2.x, or even limiting it to Hadoop.
>> >
>> > I for one would love to have a Drill that works on a single machine for
>> > scanning local files efficiently.  Kind of like AWK on steroids.
>> >
>> > On Sun, Sep 16, 2012 at 10:19 PM, Jim Donofrio <donofrio111@gmail.com
>> > >wrote:
>> >
>> > > I noticed one of the slides from the talk Tomar gave at the Google
>> meetup
>> > > mentioned using YARN for long running processes. Is the plan for Drill
>> to
>> > > only support hadoop 2.X?
>> > >
>> >
>>

Mime
View raw message