|
View previous topic :: View next topic
|
| Author |
Message |
ckchan
New User
Joined: 19 Oct 2007
Posts: 3
Location: HK
|
|
|
|
I heard from my system support that a CICS has a limitation on CPU resource (around 2xx MIPS), and also a CICS can only 'use' 1 CPU. Anyone know if it is true? and can you give me some documentation from IBM for the above issues?
Thanks a lot. |
|
| Back to top |
|
 |
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10873
Location: italy
|
|
|
|
Unless Your systems support people played some USELESS tricks in defining
the performance parameters of You MVS system both assertions are Not true!
MVS tuning is performance oriented, which means to give each process
all the resources ( CPU, STORAGE, I/O's ) needed to achieve it' s performance target.
There is not ABSOLUTELY the need to artificially limit the resources used by a
process...
if there is no resource contention, why impose limits ???
( higher priority processes will always be given more resources than a
competing lower priority one ) |
|
| Back to top |
|
|
ckchan
New User
Joined: 19 Oct 2007
Posts: 3
Location: HK
|
|
|
|
thanks for your comment.
In my understanding, system supports can 'set' the upper limit of CPU usage for batch jobs (class?), online jobs (class?) in order to have a better management of resourse(e.g. avoid batch jobs to draw to much CPU time and affect the online jobs performance).So ... is it possible to allow the CICS to have such 'setting' to bound the CPU resource?
| enrico-sorichetti wrote: |
Unless Your systems support people played some USELESS tricks in defining
the performance parameters of You MVS system both assertions are Not true!
MVS tuning is performance oriented, which means to give each process
all the resources ( CPU, STORAGE, I/O's ) needed to achieve it' s performance target.
There is not ABSOLUTELY the need to artificially limit the resources used by a
process...
if there is no resource contention, why impose limits ???
( higher priority processes will always be given more resources than a
competing lower priority one ) |
|
|
| Back to top |
|
|
enrico-sorichetti
Superior Member
Joined: 14 Mar 2007
Posts: 10873
Location: italy
|
|
|
|
Setting an upper limit, will enforce the rule also when there is no need for it !!
If the parameters are set in the right way,
and it takes a little bit of work and analysis of the performance reports,
higher IMPORTANCE processes ,
both in GOAL mode, and in compatibility mode ( ICS,IPS)
will be given the necessary resources to achieve their target response time
so ... again no need to put an artificial high water mark on resource utilization
Unless, You/your system people are mixing concepts from "LPAR capping" |
|
| Back to top |
|
|
ckchan
New User
Joined: 19 Oct 2007
Posts: 3
Location: HK
|
|
|
|
Ok. I will check with my support colleagues again. Thx.
| enrico-sorichetti wrote: |
Setting an upper limit, will enforce the rule also when there is no need for it !!
If the parameters are set in the right way,
and it takes a little bit of work and analysis of the performance reports,
higher IMPORTANCE processes ,
both in GOAL mode, and in compatibility mode ( ICS,IPS)
will be given the necessary resources to achieve their target response time
so ... again no need to put an artificial high water mark on resource utilization
Unless, You/your system people are mixing concepts from "LPAR capping" |
|
|
| Back to top |
|
|
|
|
