Skip to content


Subversion checkout URL

You can clone with
Download ZIP


Store indexes in custom data segments [moved] #958

lvca opened this Issue · 0 comments

1 participant


This is Issue 958 moved from a Google Code project.
Added by 2012-07-19T08:47:13.000Z by
Please review that bug for more context and additional comments, but update this bug.

Original labels: Type-Enhancement, Priority-Low, v1.3.0

Original description

This is follow-up request based on the discussion from

The problem under discussion was that even if one creates custom classes bound to custom physical clusters bound to custom data segments, some data (in particular, indexes) are still stored in the default data segment which might negate the benefits of having split data segments (for example, concurrent performance). 

Because the default data segment is shared, read/write access would require read/write locks which would block on concurrent access. On the other hand, if indexes were in separate segments, a write on one data segment (think - one data set) would not block reads (or writes) on another data segment (another data set).

Lvc's response to that was:
I agree that partitioning indexes in different data segments makes sense. MVRBTree API already supports this kind of partitioning allowing the selection of custom clusters. What is missed are:
* specify custom data-segments
* update the SQL create index to support this

So this issue is raised as a placeholder for potential improvement in this area.
@lvca lvca added the invalid label
@lvca lvca removed this from the 2.0rc1 milestone
@lvca lvca closed this
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.