Data point: the last time I saw very high CPU it was due to storm
object comparisons:
if A == B, when A and B are storm objects (e.g. SQLBase derived) is
very expensive; any algorithm which involves a lot of such comparisons
will lead to high python time.
e.g. if A in something
-> fine if something is a dict or set
-> trouble if something is a list or tuple
Data point: the last time I saw very high CPU it was due to storm
object comparisons:
if A == B, when A and B are storm objects (e.g. SQLBase derived) is
very expensive; any algorithm which involves a lot of such comparisons
will lead to high python time.
e.g. if A in something
-> fine if something is a dict or set
-> trouble if something is a list or tuple
-Rob