Re: Secondary Index Storage

2013-08-23 Thread Dave Brosius
the location is loaded from the cassandra.yaml, specifically the entry data_file_directories This is loaded into the class Config.java using 3rdparty library snakeyaml On 08/23/2013 06:34 AM, Nipuni Piyabasi Perera wrote: Hi, Ok, so if I want to make it separate and test, where should I

Re: Secondary Index Storage

2013-08-23 Thread Nipuni Piyabasi Perera
Hi, Ok, so if I want to make it separate and test, where should I refer? From where the keyspace refer to the location "/var/lib/cassandra/data". I have debug the code, and could see some variables (e.g jdk location etc) can be accessed from : System.getProperty(""); But this does not contain th

Re: Secondary Index Storage

2013-08-22 Thread Jonathan Ellis
Sorry, Aleksey pointed out that the index directory isn't separate from the "parent" table. Probably so that we can snapshot them both together, which is why the separate config in .yaml doesn't make sense either. On Thu, Aug 22, 2013 at 11:07 PM, Jonathan Ellis wrote: > You can already symlin

Re: Secondary Index Storage

2013-08-22 Thread Jonathan Ellis
You can already symlink the index directory anywhere you like. I'm not convinced that special-casing indexes is a good approach in general vs looking at performance of all tables. On Thu, Aug 22, 2013 at 8:16 PM, Nipuni Piyabasi Perera < nipuni880...@gmail.com> wrote: > Hi, > > I need to do a s

Secondary Index Storage

2013-08-22 Thread Nipuni Piyabasi Perera
Hi, I need to do a secondary index access improvement with a little modification to Cassandra. I could see in [1], that the performance of secondary index querying can be increased via splitting the column family data and secondary index data. Currently Cassandra saves keyspace data and secondary