Reading through example and HBase client docs I'm not expecting the below performance issue.
Does the MultiRowRangeFilter support fast forward? The docs say it does but our experience is thus far no. We are using the MultiRowRangeFilter for a set of 100 pairs of scans, no overlapping in the start and end row keys. The scan takes about 10 mins to execute and the performance is worse without a start and end row set on the scanner. A loop through the 100 pairs executing them 1 by 1 results on the complete answer in just 10s of milliseconds. It seems like a full table scan is being done without fast forward when the MultiRowRangeFilter is used. HBase 2.5.6 & 1.1.2, using the HBase Java client libraries for the queries from a web service. Regards, Jeremy
