Vijay, You can try this.
select * from my_table where <primary key> like '?%' and tid = ? (argument ? is customerid) This will do a range scan and a filter on server side. Other option is to change primary key design to (customerId int, transactionId varchar, timestamp BigInt). This will allow to query on primary key directly resulting in range scan. select * from my_table where <primary key> like '?%' (your argument should include both customerid and transactionid) Above will return all timestamps versioned records for a matching combination of customerid and transactionid. Thanks, Hemal On Fri, Jun 5, 2015 at 10:00 AM, Vijay Kukkala <[email protected]> wrote: > Cluster configuration: Phoenix 4.0.2 with Hbase 0.98 HDP 2.1 > > One of our table has a primary key (customerId int, timestamp BigInt, > transactionId varchar). > One of our use cases is to retrieve records by customerId and > transactionID. > > select * from my_table where cid = ? and tid = ? > > looking at the explain plan, it resulted in a range scan on customerID > with a filter on transactionID. > > I tried to create an index on (CustomerID, transactionID) but didn't > result any change in the explain plan. > > Trying to understand if there is a better way to handle this. In Hbase, > there is a fuzzyRowKeyFilter which allows to query on parts on the key. > > Would appreciate if someone can point me in the right direction. > thanks, > Vijay > > -- Hemal Parekh Senior Data Warehouse Architect m. 240.449.4396 [image: Bitscopic Inc] <http://bitscopic.com>
