Details
-
New Feature
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
We need to support indexing in the new platforms. This can be an umbrella issue to track the different types of indexing we'd like to support. Here's four different strategies:
1) MapReduce - support for a pure Lucene version of existing map/reduce functionality. For ultra-large bulk indexing; very high throughput, very high latency.
2) Mutations - direct mutations similar to today. very low throughput; very low latency.
3) Enqueue - ...
4) Open-> Add/Update+ -> Commit - support a begin transaction, add one or more documents, then commit (end transaction). medium throughput; medium latency.