Ravindra Pesala created CARBONDATA-1230:
-------------------------------------------
Summary: Datamap framework for Carbondata to leverage indexing
Key: CARBONDATA-1230
URL:
https://issues.apache.org/jira/browse/CARBONDATA-1230 Project: CarbonData
Issue Type: New Feature
Reporter: Ravindra Pesala
Datamap should be single point interface for indexing and pruning.
It could be two types
# 1. Coarse grained datamap.
# 2 Fine grained datamap.
h3. Coarse grained datamap
These datamaps contains the information of blocklets. so it can prune till blocklet level. It could be loaded on driver side or executor side depends on size of datamap.
Default implementation for this type is BlockletDataMap. It contains all necessary information of blocklet with stats like startkey, endkey and max and min value. Using this information all filter queries would be pruned by datamap.
h3. Fine grained datamap
These datamap contains information up to page and row level. It is stored executor side and used as part of filtering to speed up the queries.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)