Bill,
I agree its a size thing, but the ironic bit is that the creating of
these tables is usually a waste of time as the size of the zseries
complex increases.
Why? Well, this internal representation of the IO config that is
maintained by ESCM is only really used if you want to use ESCM to do
some of the more complex stuff, such as issuing commands (via ESCM) to
config chps offline across various LPARS. Its a nice concept, there's
even one command that will (in theory!) vary offline all the paths to a
director from all processors, and then take the director offline. Great,
except that as the number of LPARS increase (and the complexity of your
ioconfig), the likelyhood of one of the systems rejecting the command
also increases (almost a certainty in my experience), and then the whole
command fails... so you end up doing it manually!..Bottom line is I've
never seen any of these commands used in anger. Even smaller sites might
struggle to use these advanced commands.
If you only use ESCM to update/query the director details , ie port
name, matrix, then none of the 'internal' database built by ESCM is
really relevant. It just issues the CCWs direct to the director CUP..
So the bigger your ioconfig, the more cycles you burn at startup, and
the less likely you are of making use of the info. What a waste!
Regards
Roy
(IBM Mainframe Discussion List) wrote:
In a message dated 5/11/2006 4:42:20 A.M. Central Daylight Time,
[EMAIL PROTECTED] writes:
It still burns a
large amount of CPU at startup. From what I've seen this is due to the
size of the IO config i.e. how big the IODF is. I get the impression
that ESCM reads the IODF and builds a various internal tables and cross
matches them to available IO resources (ie paths). I'm not sure why it
takes so long & so much CPU, maybe inefficient sorting/lookup etc..
Sounds like another case of the large system effect. LPARs today in big
shops with lots of DASDs being mirrored can have 10K to 20K devices easily. And
more.
Bill Fairchild
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
