[
https://issues.apache.org/jira/browse/PHOENIX-1335?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14909528#comment-14909528
]
Andrew Purtell commented on PHOENIX-1335:
-----------------------------------------
Wouldn't it be oddly inflexible if every cell upserted on a connection is given
the same blanket TTL? It would in fact "override" any per-table and per-CF
distinctions make in schema (with below caveat). Wouldn't just about any use
case expect only to apply TTLs for some values or in some tables?
By the way a cell TTL is not a future timestamp, it's the desired lifetime of
the cell in milliseconds. Cell TTLs can't cause a cell to live beyond any TTL
defined in the table schema (due to TTL handling optimizations at the Store
level), so it would make sense to throw an exception if we detect a user is
trying to do that.
> Surface support for per cell TTLs
> ---------------------------------
>
> Key: PHOENIX-1335
> URL: https://issues.apache.org/jira/browse/PHOENIX-1335
> Project: Phoenix
> Issue Type: New Feature
> Reporter: Andrew Purtell
> Assignee: Andrew Purtell
>
> If/when per cell TTL support is committed to HBase (HBASE-10560), we should
> surface this capability in Phoenix.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
