AnsweredAssumed Answered

Scaling Approaches for a Molecule

Question asked by SatyaKomatineni3761 on May 25, 2016
Latest reply on Jun 8, 2016 by Adam Arrowsmith

Should I scale a molecule by adding multiple nodes or add more memory to the existing set of nodes?

 

Will the more memory option is the same as adding multiple nodes approximately?

**********************************************************************************************

If I set my molecule to have forked execution and say each process can be forked on a 4G JVM. I can increase the server memory to 32G allowing lot of parallel processes.

 

Say I set my flow control to "8 processes". I only have 2 nodes. I assume both nodes will be used so 2 out of 8 gets running on the 2 nodes. Will the rest 6 spawn additional processes as they can be forked? Or is my understanding wrong where processes are strictly between nodes and does not take into account spawned processes?

 

Will the spawning occur only for new distinct additional start processes?

 

If I cannot achieve 8 parallel branches of a given process using 2 nodes, should I then need 8 nodes?

 

Appreciate your help. in advance.

Thanks

Satya

Outcomes