CONNX JDBC on Linux Slower than JDBC on Windows 2008r2?

, , , , , ,
1

We have installed an instance of the connx jdbc server on a Red-Hat Linux server with 32GB of RAM and 4 CPU. I tried to benchmark the query execution time against this Linux server with another jdbc server running on Windows 2008r2 (8GB RAM and 2 CPU). These machines are all VM.

The query looks like this:

select count(*) from SUBJECT_LAST_FIRST sfl inner join Subject s on sfl.ISN_SUBJECT=s.ISN_SUBJECT
where sfl.LAST_FIRST like ‘JONES%’

I was surprised to find that the query generally run slower with the jdbc server on Linux then in Windows. The CDD used in both servers use the same ADABAS SQL Gateway and DB on the mainframe.

The difference in time seems to sit between the Begin Query Execution and the Query Execution Complete statement in the JDBC server log.

Does anybody have any idea on why the same query would run longer in Linux than in Windows? The only difference between the 2 environment is the jdbc server but not the Gateway nor the ADABAS DB.

Can someone shed some light on what transpires between the Begin Query Execution and the Query Execution complete statement often see in the jdbc server log?

May be tuning is needed on the Linux box?

Any information would be much appreciated.

Thanks.

2

Hi Min,

There are may factors to consider when it comes to performance.

  1. Network bandwidth - and # of network cards and configured speed of network cards
  2. Network latency between the data server and the JDBC server.
  3. CPU Speed, and to a lesser extent # of CPUs

I would recommend that you open a support incident for this issue, as we will need some detailed information about all of these components.