Thanks Mark,
But I have one questions about your example
(eg, you don't want processes on 2 different nodes trying to update the same chunk of data => the overhead of the internode lock manager will likely introduce an unacceptable degradation in performance).
Now I have many operations when users use some chunk of data(rows in table,keys in indexes and....) and now I have 1 working process and some processes which wait(block).
And I know, if some process use chunk of data and some processes wait it's normal, it's life.
When first process execute it's statement(end transaction) all another processes start working.
This situation will repeat in CE version or in CE lock manager enter some "new degradation in performance"?
Maybe you know still any examples?