Hello, Looks like your schema didn't make it through, but I get the idea. Phoenix expects a zero byte between the row key columns instead of a dash and there's currently no way to override this. See PHOENIX-506 for a discussion of how this could be achieved. If you're up for contributing, this would be a good one. Thanks, James
On Sun, Nov 30, 2014 at 6:00 PM, [email protected] <[email protected]> wrote: > First of all thanks for open sourcing the phoenix which is awesome. We are > definitely going to use it for our statistics on our hbase. > But since we are new to the project, we'd like to ask you a question based > on our hbase design: > > we have a legacy hbase table to store sensor data collected from a car, > and the table schema looks like the following: > > as you can see the rowkey for this table is pretty complex, it is combined > with 5 elements: signal-cartype-busnumber-messageid-timestamp. column > family is simply the value of the 'signal'. > unfortunately we are not allowed to change it at the moment. > > we want to create a view for this table and do some query based on > elements of the rowkey, can we do that in phoenix? how can we do that? > if not, could you please give us some advices on hbase table schema > optimizations? > > Best Regards > > > ------------------------------ > [email protected] >
