When using pig with HBaseStorage, pig filters should utilize hbase indexes to limit workset.

The LOAD function using HBaseStorage has filter arguments you can use limit the working set for an MR job.
e.g.
blah = LOAD 'hbase://test' using org.apache.pig.backend.hadoop.hbase.HBaseStorage('cf:field1', '-loadKey -gte foo1 -lte foo1');

It would be really great if this could also be applied to filter statements within pig, where a filter statement within pig e.g.
blah2 = FILTER blah by key=foo1; or
blah2 = FILTER blah by key > foo1 and key < foo2;

would actually limit what is retrieved from hbase, so big has a smaller working set to perform MR on.