Can you provide a very simple, but indicative, example of what you are looking for? Maybe a few rows and details of what you would be looking for. As you already know, the filter language documented at https://hbase.apache.org/book.html#thrift.filter_language will end up scanning everything. And also as you know, creating another table whose rowkey is aligned with your query you want to run fast will take effort to keep in sync as you upsert your data. There's always Phoenix is you want to stay completely in HBase, but maybe a hybrid of HBase and Solr might work. Again, my a simplified example could help.