Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Drill, mail # dev - Tajo


Copy link to this message
-
Re: Tajo
Julian Hyde 2013-02-26, 21:59
I'd also urge the various groups to focus on building components/libraries, not whole systems. Components are re-usable if their interfaces are well designed. Open source is a force multiplier, but we lose a lot of that advantage if we don't design for re-use.

On Feb 26, 2013, at 1:30 PM, Jacques Nadeau <[EMAIL PROTECTED]> wrote:

> I completely agree.  The more wood, the better. I've reached out to
> Hortonworks guys and am doing an in person preso to their team.  Hopefully
> that will help facilitate collaboration opportunities on Tez/Stinger and
> Drill.  I also am working on connecting with the Tajo guys to see what
> synergy is possible.
>
> To me, the biggest gap is going to be around differences in vision and
> execution strategy.  It seems like many of the systems being built or
> discussed tend to either defer or ignore some fairly important market or
> research trends.  Things like complex schema-lite data and columnar
> execution come to mind when looking at Tajo--there are others when looking
> at other projects.  Typically, the response is that they'll do these things
> later.  I applaud that response but question the feasibility.  These things
> often have fairly substantial underlying architectural implications that
> will be difficult to "fit-in" later.
>
> On the flipside, we are all resource constrained and the last thing that
> everybody should be doing is reinventing the same wheel. I hope that we can
> find places to share work.
>
> J
>
>
> On Tue, Feb 26, 2013 at 10:49 AM, Christopher Merrick <
> [EMAIL PROTECTED]> wrote:
>
>> Is there any reason why we shouldn't reach out and see if there's an
>> opportunity to join forces?  Between Drill, Impala, and the Hortonworks
>> Stinger initiative <http://hortonworks.com/blog/100x-faster-hive/> that I
>> just read about, there are plenty of different teams working on this
>> problem.  If their vision is really that similar to ours, I think it makes
>> sense to put more wood behind one arrow.
>>
>>
>> On Tue, Feb 26, 2013 at 1:09 PM, Ted Dunning <[EMAIL PROTECTED]>
>> wrote:
>>
>>> Yeah... I don't see much difference either.
>>>
>>> I read their slides and they could be giving a Drill pitch for all I
>> could
>>> tell.
>>>
>>> On Tue, Feb 26, 2013 at 12:31 AM, Timothy Chen <[EMAIL PROTECTED]>
>> wrote:
>>>
>>>> Not sure folks seen this?
>>>>
>>>> http://wiki.apache.org/incubator/TajoProposal
>>>>
>>>> It also mentioned about Drill and how they claim to be different.
>>>>
>>>> Tim
>>>>
>>>
>>