incubator-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Billie J Rinaldi <billie.j.rina...@ugov.gov>
Subject Re: Accumulo incubator proposal: Statement of Concern
Date Wed, 07 Sep 2011 19:47:37 GMT
On Wednesday, September 7, 2011 1:34:20 PM, Stack <stack@duboce.net> wrote:

> I agree w/ Doug that 'unlikely to' is not a correct characterization.  

Would the following alteration be more accurate?
"It may be possible to incorporate the desired features of Accumulo into HBase.  However,
the amount of work required at the current time would slow development of HBase and Accumulo
considerably."

> But rumor has it though that the differences while small looking when
> described in a short incubator proposal, in implementation, the code
> is very different making an integration project, unfortunately, a
> piece of work.

Yes, the implementation is very different, and we had difficulty capturing that in the proposal.

> hbase TRUNK coprocessors seem to be a more generic Iterator facility

Some types of functions (e.g. query-time aggregation) can be implemented in both coprocessors
and iterators, but coprocessors will not easily support the entirety of iterator functionality.
 Nor is the reverse true.  The two models present different programming mechanisms for server-side
processing.  It would be useful to have both in the same project.

Also, I saw that you wondered in a different forum whether editing locality groups requires
taking a table offline.  The table can remain online because files contain information about
their own column groupings.  Thus the changes can take place as new files are written to disk
and old files are compacted.

Billie

---------------------------------------------------------------------
To unsubscribe, e-mail: general-unsubscribe@incubator.apache.org
For additional commands, e-mail: general-help@incubator.apache.org


Mime
View raw message