[
https://issues.apache.org/jira/browse/CASSANDRA-6477?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14540775#comment-14540775
]
Carl Yeksigian commented on CASSANDRA-6477:
-------------------------------------------
MV allow us to route queries for high-cardinality non-partition columns so that
they are single node queries rather than having to fan-out on read. The reason
this was renamed is that we want to prevent a fan-out after we get back the set
of keys, so we store the data that we want to query with the keys, thus
creating something closer to MV than global indexes. In addition, we want to
make sure that users understand why they can't select arbitrary columns from
the table; they will only be able to use the columns they specify when creating
the MV.
We also want to take something that is done by many different users currently,
which is denormalizing the data into new tables, and bring that in so that we
can make sure that the edge cases are handled properly, and make it easier for
users to model their data.
bq. Is it fair to say that the primary technique for using this feature is to
have one base table and n views of that table
Yes, as many tables as the users currently create and use themselves, they
should be able to create materialized views for.
bq. Would it also be sensible to select a subset of rows?
This won't be in the initial version, but it is something that we will probably
add on later, along with richer select statement support
> Materialized Views (was: Global Indexes)
> ----------------------------------------
>
> Key: CASSANDRA-6477
> URL: https://issues.apache.org/jira/browse/CASSANDRA-6477
> Project: Cassandra
> Issue Type: New Feature
> Components: API, Core
> Reporter: Jonathan Ellis
> Assignee: Carl Yeksigian
> Labels: cql
> Fix For: 3.x
>
>
> Local indexes are suitable for low-cardinality data, where spreading the
> index across the cluster is a Good Thing. However, for high-cardinality
> data, local indexes require querying most nodes in the cluster even if only a
> handful of rows is returned.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
