Hmmm.  Each index build will require a potentially resource intensive sort
- it's not obvious to me that this would actually be better.  If you load
after creating indexes you're able to do this, and it's an interesting Q
whether a pipeline is better or not....  I think we actually chose not to
do this due to the resource implications.  But - open Q.  I'd put it down
on the wish list, though - online single index construction would be higher
on the list, for instance, IMO.
On Sep 26, 2015 3:26 AM, "abdullah alamoudi" <[email protected]> wrote:

> Hi AsterixDB Dev,
>
> Some data management systems provide an option for users to defer index
> build operations. After the user defines multiple indexes, they can kick
> off the index build process for multiple indexes at once, effectively
> saving multiple dataset scan cost.
>
> We don't and it seemed like an easy and useful thing to do. So, should we
> create a JIRA for this feature?
>
> Amoudi, Abdullah.
>

Reply via email to