[
https://issues.apache.org/jira/browse/JENA-782?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14127065#comment-14127065
]
Andy Seaborne edited comment on JENA-782 at 9/9/14 3:01 PM:
------------------------------------------------------------
The tuneable parameters are:
# File mode (direct or mapped)
# Block read cache size (only meaningful for direct mode)
# Bock write cache size (only meaningful for direct mode)
# Node to NodeId cache size
# NodeId 2 Node cache size
# Node miss cache (nodes known not to be in the node table).
The creation time parameters could be:
# Blocks size
# Triples indexes
# Quads index
If a required access pattern needs an index that does not exist, a (partial)
scan of the nearest matching index is done so as long as there is one index for
each of triples and quads, functionality if not lost. Having just SPO, and SPOG
would make for a compact database good for URI deferencing.
was (Author: andy.seaborne):
The tuneable parameters are:
# File mode (direct or mapped)
# Block read cache size (only meaningful for direct mode)
# Bock write cache size (only meaningful for direct mode)
# Node to NodeId cache size
# NodeId 2 Node cache size
# Node miss cache (nodes known not to be in the node table).
The creation time parameters could be:
# Blocks ize
# Triples indexes
# Quads index
If a required access pattern needs an index that does not exist, a (partial)
scan of the nearest matching index is done so as long as there is one index for
each of triples and quads, functionality if not lost. Having just SPO, and SPOG
would make for a compact database good for URI deferencing.
> Provide configurability of TDB in databsse setup
> ------------------------------------------------
>
> Key: JENA-782
> URL: https://issues.apache.org/jira/browse/JENA-782
> Project: Apache Jena
> Issue Type: Improvement
> Components: TDB
> Affects Versions: Jena 2.12.0
> Reporter: Andy Seaborne
> Assignee: Andy Seaborne
>
> Creating TDB databases and attaching to existsing ones is quite fixed
> currently. Any internal paramaters are quite difficult to manipulate and
> they are globals.
> Configurability of TDB databases has two dimensions:
> # Initial setup (e.g. block size)
> # Per JVM (e.g. cache sizes).
> (The file mode is actually a per JVM feature but should probably be regarded
> as an initial setup feature only.)
> With configurability, TDB and Fuseki could be scaled down for small databases
> and for many small databases in one JVM.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
