A month or so ago the issue of Type1 versus Type2 PDSEs was brought to the attention of this board in terms of space reuse and performance, etc. etc.
Personally, I'm sold on using PDSE Type 2 but now the question:
I have an installation that just migrated to PDSE Type 1 that involves a large number of datasets. Is it worth the effort to (re)convert them all to PDSE Type 2? If yes, where should we look to see benefits? The tests run to-date show marginal if any benefits but I'm not convinced we're looking in the right places for those benefits.
Alternatively, a re-engineering effort is going to be re-allocating those datasets "in the future" with "in the future" dates yet to be determined as there are a variety of other changes taking place at that time as well.
Would you wait till the re-engineering effort (long-term) or convert them "now"?
Your opinion is of great interest!
Only datasets that are permanently enqueued would benefit.
So linklisted and any files allocated to active/active started tasks
Hmmm... ok.... so files such as load libraries for CICS or IMS/DC regions?
Yep, if they are always up.
Linklist was the killer one for us in a parallel sysplex because that very rarely comes down entirely.
See what road blocks you have and try to get rid of them.
IBM strategic direction is for LIBRARY type.
Ultimately, IBM will drop the old format in order to make libraries work with guest operating systems.
Very long term systems architecture should take this into consideration.
IBM products use LIBRARY. This should tell you something.
The link to the IBM documentation on this is:
IBM Knowledge Center
As you can see in the doc they say type 2 performance is better than type 1 with one exception:
So you this would indicate that LOADLIBs with large programs can have slower performance with type 2 PDSE instead of type 1.