Jason Kania created CASSANDRA-11319:
---------------------------------------
Summary: SELECT DISTINCT Should allow filtering by where clause to
support time series
Key: CASSANDRA-11319
URL: https://issues.apache.org/jira/browse/CASSANDRA-11319
Project: Cassandra
Issue Type: Improvement
Components: CQL
Environment: Cassandra 3.0.3
Reporter: Jason Kania
Due to the lack of built in sharding, we have been trying to split very wide
rows. However in trying to find all the sharding column values after the fact,
we have not been able to find a solution that is manageable. If a table is
defined as follows:
CREATE TABLE IF NOT EXISTS "sensorReadings"
(
"measurementList" blob,
"sensorId" int,
"sensorUnitId" int,
"shardId" int,
"time" timestamp,
PRIMARY KEY ( ("sensorUnitId", "sensorId", "shardId"), "time" )
);
then
select DISTINCT "sensorUnitId","sensorId","shardId" from "sensorReadings";
will give all the unique partition keys but this can still be a very large set
and so it should be possible to refine this with a where clause that contains
only partition columns ie:
select DISTINCT "sensorUnitId","sensorId","shardId" from "sensorReadings" WHERE
"sensorUnitId"='sensor17' AND "sensorId"=8;
Without this ability, we am forced to keep a table with available shardIds and
update on every write so that we can even query the original table. While
several scenarios allow the shardId to be determined automatically, attempts to
iterate over the shards are seriously hampered.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
