Limiting resource use during maintenance operations.

[Marcin Januszkiewicz <ka...@gmail.com>]

Hi,

I have a 5-node HBase cluster with around 12GB of RAM available to
RegionServers. I've been evaluating Phoenix with a ~20M row HBase
table with 16 regions, and have had some problems. Deleting an index
from a view will quickly cause an out-of-memory error on the
RegionServers. Creating the index works reasonably fine. When I try to
create a table from the hbase table instead of a view, I have the same
problem - the RegionServers eventually crash from the load.

Is there a way to control the resources that will be available to
Phoenix to execute these actions? Is there a recommended minimum
hardware setup? Will there be tools available similar to the async
table indexer?

Regards,
Marcin