Piotr Pastuszka created CASSANDRA-15106:
-------------------------------------------
Summary: Fat Cassandra and CQL nearer to SQL
Key: CASSANDRA-15106
URL: https://issues.apache.org/jira/browse/CASSANDRA-15106
Project: Cassandra
Issue Type: New Feature
Components: Legacy/CQL
Reporter: Piotr Pastuszka
I understand that Cassandra is a dispersed lightweight, performance /
scalability / reliability database. For this reason, it has limitations not to
overload the single node.
But these restrictions for someone who administers the database or explains the
problems are too big. They remind us of the attempt to use mono monitors in the
era of color monitors.
At the beginning I worked on Cassandra 2.0. At that time it was not possible
to use IN,>, <. Also, searching for several partitions was not possible.
Now in 3.11.4 it's all there is, but there are several necessary
functionalities that are very missing.
At the moment, to make the functions available, you need to change the
parameter in the configuration.
You could add below features and parameters to enable:
a) sorting on Cassandra's side (this could be done by the coordinator): ..
order by xxx. Sorting mechanisms are available in Cassandra because data is
sorted in partitions and there are indexes
b) subqueries. CASSANDRA-8846 SubQuery and IN operator implementation
c) CTAS - CREATE TABLE AS SELECT - CASSANDRA-8234 and INSERT INTO .. SELECT ...
d) DBLINK to other databases (Oracle, MSsql)
e) functions in which the CQL query can be called and return a result
f) external table (file on disk visible as a table under database).
I know there is SPARK, but it's a programming solution, not for administrators
or testers.
I am the administrator of Oracle and Cassandra.
I know SQL quite well and working with CQL is a torment.
The CQL editor is _very_ good, but CQL restrictions make it difficult to solve
any problem.
I can create all points a) -f) as independent request, but as I can see, all
such ideas are rejected / postponed.
The point is, however, to change the approach. Today's huge Cassandra's
restrictions reject many potential users.
Maybe you can combine Cassandara and Spark and give additional functionality to
CQL, or do it only on Cassandra side.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
