drill-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From AnilKumar B <akumarb2...@gmail.com>
Subject Re: Cassandra storage plugin
Date Wed, 17 Jan 2018 22:04:17 GMT
Hi Matt,

>> I note there are some old Jira issues about Cassandra storage, and have
this concept as to why it could be very valuable for Drill. Can anyone
support or refute the idea?
I agree, this will be a very valuable feature for Drill.

>> Cassandra is an excellent engine for high volume ingest, but support for
aggregations and scans is very limited. Would a Drill plugin support the
ability to query a Cassandra keyspace using the partition keys, then
perform aggregations within Drill, adding an "analytics layer" in Drill
onto Cassandra?
Yes, this should be possible from Drill.

>> This seems like a very useful approach, but I don't know the level of
effort in getting Calcite to properly determine which part of the SQL query
passed to Drill can be segmented to CQL to push down to Cassandra, and then
aggregate to summarize the results.
Yash, already done some work on this feature and there are first round of
review comments. But currently there is no active effort on this.
To resume, on this feature,  good reference will be Spark cassandra
connector https://github.com/datastax/spark-cassandra-connector.


Thanks & Regards,
B Anil Kumar.

On Tue, Jan 16, 2018 at 12:21 PM, Matt <bsg075@gmail.com> wrote:

> I note there are some old Jira issues about Cassandra storage, and have
> this concept as to why it could be very valuable for Drill. Can anyone
> support or refute the idea?
>
> Cassandra is an excellent engine for high volume ingest, but support for
> aggregations and scans is very limited. Would a Drill plugin support the
> ability to query a Cassandra keyspace using the partition keys, then
> perform aggregations within Drill, adding an "analytics layer" in Drill
> onto Cassandra?
>
> This seems like a very useful approach, but I don't know the level of
> effort in getting Calcite to properly determine which part of the SQL query
> passed to Drill can be segmented to CQL to push down to Cassandra, and then
> aggregate to summarize the results.
>
> Is there any active effort in this area?
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message