Ed,
MDC is almost always faster than disk cache. First of all, the system
doesn't have to initiate an I/O on your behalf to fetch data from MDC. If
it's not in MDC, it does have to do an I/O, even if the data is found in
controller cache.
If the data happens to be in both MDC and
David,
What kind of SHARE statements do you have in the Linux servers' directory
entries. That's what determines how much of the available CPU resource the
users get.
Marty
Martin Zimelis
Principal
maz/Consultancy
_
From: The IBM z/VM
Rob has answered your second question quite well.
As to the first one -- whether to use QUICKDSP for Linux guests -- you first
need to know what QUICKDSP does. It overrides the CP Scheduler's decisions
on whether to admit a virtual machine to the Dispatch List (contending for
resources) or to
Sir Jeff, the Builder of Bridges,
As one who not long ago took the same path, I can speak highly of the
benefits of being the boss of one's own time (at least, as far as the dog
will let me). I'm not sure that 26+ years in the same place qualifies you
to sing the Song of the Gypsy Programmers.
Actually, since the announcement was after the close on the NYSE yesterday,
it wasn't down. In fact, as of noon EST today, the stock is up $7.31.
Marty
-Original Message-
From: The IBM z/VM Operating System
[mailto:ib...@listserv.uark.edu] On
Certainly, there will be a limit eventually. What you ought to be asking to
determine if you're approaching the limit is
1. Is the response time (time it takes to serve up data) significantly
different than when there was less data in the data store? (This implies
that you've been measuring
Try the Locate subcommand. And welcome to the fold.
Marty
Martin Zimelis
Principal
maz/Consultancy
_
From: The IBM z/VM Operating System [mailto:ib...@listserv.uark.edu] On
Behalf Of David Logan
Sent: Friday, December 19, 2008 9:42 PM
Richard,
Nothing unusual will happen at first. As the mod 3s start filling up,
the system will attempt to page preferentially to the mod 9s because their
performance will be better. (The paging subsystem will be able to write
longer blocks of pages to the less full volumes.) In the extreme,
I think there may be different interpretations of what deleting an entry
in the CP_OWNED list means. I *think* Richard Corak means replacing a
VOLSER entry with the Reserved keyword, _not_ compressing the list after
deleting an entry.
The pointers to SPOOL files from the Warm Start area and
I think you're asking the wrong question. The right question is probably
why was it CPOWNed on the prior releases? If there's no good reason, then
it didn't have to be in the CPOWNED list in the prior releases and doesn't
have to be in 5.4. (With enough research, you may discover that the
In your LOCATE stages, the Rexx variable GUEST must be outside of the quotes
so value substitution happens.
Marty
Martin Zimelis
Principal
maz/Consultancy
_
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of
Look at
SET SHARE guest ABSOLUTE nn LIMITHARD nn
where nn is some value less than 100 that gives you enough breathing
room for other guests to get a shot at the processors. In the longer term,
it sounds like you've defined this guest with 5 (or more) virtual CPUs. You
might want to give
Was I the only one who followed the link in Phil Tully's message of about
three hours ago?
http://en.wikipedia.org/wiki/Illegitimi_non_carborundum
Marty
Martin Zimelis
Principal
maz/Consultancy
_
From: The IBM z/VM Operating System
IIRC, pages read from SPOOL-resident DCSSs are counted as paging I/O, so the
scenario you describe in your third paragraph is certainly a possibility.
Marty
-Original Message-
From: The IBM z/VM Operating System
[mailto:[EMAIL PROTECTED] On
might look up Al Williams (Sir Al the Unflappable) for more information.
Marty Zimelis (Sir Marty the Preceptor)
_
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Mary Anne Matyaz
Sent: Wednesday, May 07, 2008 10:09 AM
To: IBMVM
Martha,
My recollection is that there's only an allocation bit map on a
page-extent basis. The only way I can think of to track down the page is to
figure out what the CCPV value* is for the page, then traverse all the
PGMBKs for the suspect users, looking for a matching DASD slot address in
See today (12 Feb 2008)'s Dilbert at http://www.dilbert.com (or, if it's no
longer the 12th of February, see
http://www.dilbert.com/comics/dilbert/archive/dilbert-20080212.html).
Marty
Martin Zimelis
Principal
maz/Consultancy
James is probably correct. CMS is aware enough of its environment to ask CP
how much storage it has, thus never has to touch each of its pages (as some
operating systems do). It's also better behaved (for a virtual storage
environment) in releasing pages it isn't using.
For a CMS-intensive
Bob,
Right name, but I believe the wrong derivation. The 67 in CP-67 comes
form the fact that it ran on the S/360 model 67, the only production model
of the S/360 line that implemented Dynamic Address Translation (DAT) --
virtual storage.
Some would argue that was the first version of VM.
Almost, but not quite. The RC=3 means invalid option. The proper form of
the query is CP QUERY USER userid, which will report not logged on for a
user that doesn't exist. Spooling your punch (after querying it to capture
the existing setting) to the target user will return RC=53 if the user
Lionel,
Old news? You really ought to read this stuff before forwarding links
to it. In the very first paragraph (after the intro), they show a line-mode
log on to VM/370 and make reference to 43xx and 30xx processors. This
material is 30+ years old.
Marty
_
Spool records (and page records and warm start and checkpoint) are addressed
as CCPV:
CC - cylinder number on the volume (2 bytes)
P - page number within cylinder (1 byte)
V - volume number - index number into the CPOWN list (1 byte)
The last one is why you can't move a
- Hourly rate: $40-70
That's an interesting number. A complete fantasy, but an interesting
number all the same.
For $40/hr, what do they expect to get? Remote support?
Please, let's not start this discussion... again.
Nah! It's Marty's Fifth Rule of Performance: When in doubt, blame it on
network delays.
-Original Message-
From: The IBM z/VM Operating System
[mailto:[EMAIL PROTECTED] On Behalf Of Marcy Cortes
Sent: Wednesday, August 01, 2007 2:49 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: FW:
There's a treasure trove (including that paper) at
http://vm.marist.edu/~pipeline
Marty
_
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Lionel B. Dyck
Sent: Friday, July 27, 2007 4:18 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: pipe
Lionel,
My first rule of thumb is Everyone's thumb is a different size. That
is, what works for me isn't necessarily going to work for you. Using
someone else's experience will work for you right up until it doesn't.
I'd suggest going through the SHARE Proceedings archives and/or the
As long as you're in PIPES, take a look at the example for DEBLOCK NETDATA
in the author's help (PIPE AHELP DEBLOCK).
Marty
Martin Zimelis
Principal
maz/Consultancy
_
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf
You are misinterpreting the usage note. What it means is that relative
shares are doled out as integers. In other words, while a virtual machine's
relative share is normally divided equally amongst its vCPUs, when you get
to the case where the number of those vCPUs exceeds the total relative
] On
Behalf Of Marty Zimelis
Sent: Friday, March 23, 2007 3:16 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: CPU usage -- virtual or dedicated ?
You are misinterpreting the usage note. What it means is that relative
shares are doled out as integers. In other words, while a virtual machine's
relative
I found the same info that Miguel did. A Google search on the street
address shows that the building contains what appear to be several tenants
including a stock broker, an ISP, and some offices of the City University of
London.
_
From: The IBM z/VM Operating System [mailto:[EMAIL
I'm guessing you didn't give us the full request. I assume that records
without the string APPLES pass unchanged. Records with the string APPLES:
do you want to replace the string or the entire record with ORANGES? Or do
you want to insert a new record (ORANGES)? If an insert, before or after
pipe insert ?
Marty,
Yes I want to pass the entire file to 'new file a' and insert ORANGES before
APPLES.
Thanks
-Original Message-
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED]
Behalf Of Marty Zimelis
Sent: Tuesday, March 06, 2007 11:41 AM
To: IBMVM@LISTSERV.UARK.EDU
Modify the SPECS stage to
'SPECS /ORANGES/ 1 write 1-* 1 |',
emiting the string ORANGES before copying the existing record, rather than
after it.
Marty
_
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Huegel, Thomas
Sent:
people can do with an entire programming language. Be careful who
you mock. Many of the regular contributors to this list have been invested
Knight Members of the Order and are proud of the honor. :-)
___
Marty Zimelis
Sir Marty the Preceptor
Order
Richard,
What environment is this pipe running in? Clearly, DELETE USER is not a
native CMS command.
Marty
_
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Schuh, Richard
Sent: Thursday, December 14, 2006 4:09 PM
To:
Hi Dave,
First, I don't think IBM is branching out into the healthcare area. (One
assumes you meant patent infringement.)
Second, I'm with Richard Schuh. I can't get the URL to work either. And
yes, I noticed that it wrapped. Even pasting the two bits together doesn't
yield a working
Tom,
My approach would be a bit different from Colin's because I'm assuming
that the number of terms inside parentheses is unknown.
'*: |',
'split before /(/ |',
'split after /)/ |',
'b: between /(/ 1 |',
'change /,/ xff |',/* or any other character not in data */
'f:
I'd reiterate what David Kreuter said: With sufficient storage and a
reasonably well-configured I/O subsystem, you should be able to sustain 90%
or greater CPU busy without impacting guest response time. Note that those
are not trivial provisos.
Marty
Steve,
It sounds like you're a little confused about
what Q XSTORE [MAP] reports. In your current situation, you correctly
point out that no XSTORE is attached to any user. What this means is that
it's all available to CP for its two uses: primary page space and minidisk cache
(MDC).
ur system,
these users have a USER DIRECTORY with a MACHINE TYPE XC
Steve
Marty Zimelis
[EMAIL PROTECTED] Sent by: The IBM z/VM Operating System
IBMVM@LISTSERV.UARK.EDU
10/24/2006 10:03 AM Please respond to The IBM
I can't speak to the purchase price of core memory. But I believe most
S/360 and S/370 mainframes were leased in those days. I have a vague
recollection (circa 1969) of core memory leasing for $7,500/month for 256KB.
Someone who remembers what the lease:purchase ratio was will have to do the
Hey Scully,
There may be a lot of confusion, but there's only one correct answer: It
depends. :-) Seriously, the SHARE value in the directory (or set by
command) is a function of the virtual machine, not the virtual processor. A
four way with a SHARE of Relative 100 gets a relative 25
-Original Message-
From: The IBM z/VM Operating System
[mailto:[EMAIL PROTECTED] On Behalf Of Scully, William P
Sent: Tuesday, September 26, 2006 3:25 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: MIPS vs. Virtual CPUs
Let me refine this question.
- Two users. Similar workload.
-Original Message-
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Schuh, Richard
Sent: Wednesday, August 09, 2006 4:49 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: External Page Usage
With our current configuration - 36GB main, 26GB XSTORE, 19 3390-03s for
paging,
Brian,
To the best of my knowledge, you're stuck with the structure you
described in your note. The syntax of label definitions and references
requires it. If you're concerned about the number of lines added to your
pipeline with each selection stage, note that since the pipeline description
, look at
the LOOKUP stage of Pipelines with the AUTOADD parameter as the basis of a
possible solution that doesn't require sorting the file.(Though,
depending on your answer to the question about the disposition of duplicates,
that might not work either).
Marty Zimelis
From: The IBM z
-Original Message-
From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On
Behalf Of Richard Schuh
Sent: Saturday, June 10, 2006 3:53 PM
To: IBMVM@LISTSERV.UARK.EDU
Subject: Re: the IND LOAD command
Steve Gentry wrote:
On VM 4.3 when the IND LOAD was issued it returned,
I'd be inclined to go with Rich's suggestion. While it's true that your
z/VM isn't paging much now, there will almost certainly come a time when it
will. Rather than suffering erratic response time due to paging directly to
DASD (even the Shark's cache is on the wrong end of the channel cable)
48 matches
Mail list logo