Hi Vladimir,
By default, Phoenix queries show you the "latest" data. You can override
this at connection time and do "flashback" queries, DDL, and DML as
described here:
http://phoenix.incubator.apache.org/faq.html#Can_phoenix_work_on_tables_with_arbitrary_timestamp_as_flexible_as_HBase_API

What ever optimizations HBase does, Phoenix will see too, since at the end
of the day, we're issuing a regular set of scans.

Is this the kind of functionality for which you're looking? Or is it more
along the lines of a query returning multiple versions of the same row:
https://issues.apache.org/jira/browse/PHOENIX-590? This isn't supported yet.

Or maybe something else?

Thanks,
James


On Sat, Apr 5, 2014 at 10:42 AM, Vladimir Rodionov
<[email protected]>wrote:

> I am not sure it was implemented already, or it was not  ...
> It would be nice to have a way to tell Phoenix which column is naturally
> mapped to HBase timestamp. This will greatly improve range queries on
> this column by utilizing internal HBase optimizations (skipping store
> files, which are completely out of the specified timestamp range).
>
>
> -Vladimir Rodionov
>

Reply via email to