phoenix-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rohit Jain <rohit.j...@esgyn.com>
Subject RE: Drill on Phoenix
Date Thu, 16 Jun 2016 17:29:27 GMT
Tom,



One of the issues with JDBC connectivity is that there is no way to have
parallel connections between the source and the target via a simple JDBC
connection.  So if you have very few rows returning, JDBC is fine.
However, if there is a large amount of data where you would need
parallelism, there is no way to get parallel connections between a source
storage engine and a target query engine.



I think we need a change to the JDBC standard to allow for this.  It would
be complicated, but in this world of big data where everything runs in
parallel, it would seem that this would be a worthwhile effort by the
community.  We need the ability to exchange partitioning information for
both ends (how data is partitioned in the storage engine and how the query
engine would like it partitioned – hopefully they are aligned), repartition
if necessary, float parallel threads automatically to accomplish this
(should be configurable), and support this for both selects and
inserts/updates, including rowset support for efficient buffering, …



Rohit



*From:* Tom Barber [mailto:tom.barber@meteorite.bi]
*Sent:* Thursday, June 16, 2016 11:41 AM
*To:* user@drill.apache.org
*Cc:* user@phoenix.apache.org
*Subject:* Re: Drill on Phoenix



you should be able to use the jdbc connectivity to connect like other jdbc
datasources

On 16 Jun 2016 17:35, "Alex Kamil" <alex.kamil@gmail.com> wrote:

Can Drill be integrated with Apache Phoenix?

Thanks
Alex

Mime
View raw message