bloomberg / comdb2

Bloomberg's distributed RDBMS
Other
1.39k stars 214 forks source link

Optimizing comdb2 for running TPCH (OLAP) benchmark #587

Open QinyuHu opened 6 years ago

QinyuHu commented 6 years ago

Same as TPCC, I also got a question about how to tune parameters for TPCH. But since this is an OLAP database, I am not sure which part I should focus on. We are using TRANSACTION_READ_COMMITTED isolation but since this is OLAP so we will only run it with 1 terminal. (I don't think the scale factor matters here; please correct me if I am wrong.)

I am not sure from which part should I begin to think about this and which part should be same for both OLAP and OLTP.

Thanks!

riverszhang89 commented 6 years ago

You could start with this thread - those are all general optimization tips.

QinyuHu commented 6 years ago

👍 thanks :) Also there is one question I got confused. In TPCH benchmark, there is a value we need to set for generating data. But I didn't find anything for Comdb2. Do you have some suggestions for which might be chose for Comdb2? Currently I am using ORACLE but not sure this will create some issues or not.

# Current values for DATABASE are: INFORMIX, DB2, TDAT (Teradata)
#                                  SQLSERVER, SYBASE, ORACLE, VECTORWISE
riverszhang89 commented 6 years ago

Your benchmarking software does not support Comdb2 yet. You could either try to write a dialect/extension/configuration for the software, if possible; Or go with any database in that list - The difference in DDL/DML support by different database vendors should be minor. Either way, I'm afraid there's nothing we can do about this.