You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A tweak for jobs with "heavy" components and low-record count...
Right now we are concurrent on a record-by-record level. This is great for most batch jobs since it does a good job at automatically spreading the load evenly on processors and avoids bottlenecks etc.
But say you have just one record which needs to do a couple of (independent) steps. For instance, it might be calling 3 different web services (transformers), without any particular dependency/chaining inbetween. Then the 3 transformers should be invoked in parallel and thereby the overhead of waiting for the previous step to finish be gone.
The text was updated successfully, but these errors were encountered:
A tweak for jobs with "heavy" components and low-record count...
Right now we are concurrent on a record-by-record level. This is great for most batch jobs since it does a good job at automatically spreading the load evenly on processors and avoids bottlenecks etc.
But say you have just one record which needs to do a couple of (independent) steps. For instance, it might be calling 3 different web services (transformers), without any particular dependency/chaining inbetween. Then the 3 transformers should be invoked in parallel and thereby the overhead of waiting for the previous step to finish be gone.
The text was updated successfully, but these errors were encountered: