Does anyone else have 10,000 elements in a single system?

[SBAshby]

All my JCL components are in the same system and I wondered if anyone else had this same situation.

 

Stuart

[John Dueckman]

Hi Stuart!

 

Do you mean 10,000 elements in a single system spread across different types? Or 10,000 elements of a single type within a single system?

[SBAshby]

I have both scenarios

[John Dueckman]

My experience is that the first one (10K elements in a single system) is relatively common and the second circumstance (10K elements in a single type within a single system) isn't unusual.

 

But that said, your description indicated all your JCL (regardless of which business application it's associated to) is all within one Endevor system. My own experience and opinion is that doesn't serve the developers particularly well; it tends to help them more if the JCL is kept with the other bits of the defined Endevor business system. Although, again that said, my experience is Endevor provides a logical view of a system JCL library; Endevor maintained the JCL element in its base/delta library and the final MOVE processor dropped a copy into what we were commonly using (a library named SYS2.JCLLIB.... which contained the composite of all JCL for all systems).

[Phontilla Shuffitt]

I agree with John, although make sure you have some sort of element checking to make sure the single library being updated by multiple systems/subsystems do not get overlaid by mistake.

[SBAshby]

Thank you for the confirmation John.Dueckman & Phon_Shuffitt

 

I was asking because I seem to be the only customer with a performance issue and I wanted to understand if I am unique or not.

 

Stuart

[John Dueckman]

Now you've piqued my interest on a different subject! What kind of performance issue are you having? During MOVE? GENERATE? Package processing? Is your suspicion the number of elements in the type? The type of BASE library? Delta?

 

There are all sorts of things one can do to tweak performance....

[SBAshby]

LIST processing is giving the symptoms under ISPF, batch, and API (CM EW/Eclipse)

 

It appears to happen when a system contains more than 10,000 elements and a query is issued for a type with no element specified in a multiway parallel sysplex and either RLS or LSERV (but not native VSAM) are used.

 

I am happy that CA Support are on to it but I wondered if this is a unique set of conditions as it hasn't been raised before.

 

Stuart

[John Dueckman]

Interesting. I will be curious as to the "cause" as well.

 

Is element the only variable not specified? As in "LIST ELEMENT * WHERE ENV 'X' SYS 'A' SUB 'B' TYPE 'JCL' STAGE 'D' OPTION SEARCH"? Or are you "wildcarding" the other variables as well (except TYPE)?

 

It makes sense that the more you wildcard, the more cycles Endevor has to spend.... but it still needs to be "realistic"...

[SBAshby]

Hi John,

 

Your example SCL is exactly what we are doing. No wildcards just ELEMENT *