On Mon, 21 Apr 2008 12:54:54 +0200, Barbara Nitz [EMAIL PROTECTED]
wrote:
How fast can it be if well tuned and configured and with the best
hardware options?
FAST.
There is no canned answer
Our average GRS structure response time is less than 0.008ms. But that's
GRS (almost no data
Ron,
From the numbers you quote it seems it is possible for XCF to
significantly outperform TCPIP, which is the question I was asking, so it
is going to be worth my company's time to investigage further why our XCF
response times are so poor.
If you're running a sysplex, it is always a good
On Tue, 22 Apr 2008 13:21:44 +0200, Barbara Nitz [EMAIL PROTECTED]
wrote:
If you're running a sysplex, it is always a good idea to tune XCF (as far as
that goes), but keep in mind that my response time numbers are for the
ISGLOCK structure. We monitor that because it transfers almost no data
Guys,
Thanks for the input. I'm sure our system is not optimal, we've only
just
started to play with it.
The question I'm really asking, but obviously didn't make clear enough, is -
How fast can it be if well tuned and configured and with the best hardware
options?
If it's not going
How fast can it be if well tuned and configured and with the best
hardware options?
FAST.
If it's not going to be faster than TCPIP, i.e. turn around times of less
than a milliSecond, then it has no advantage over TCPIP and has the
drawback that it doesn't work to non-mainframes.
We need to
Is there any point to having XCF communications between LPARs or would
TCPIP do just as well for that?
You must have XCF for Parallel SYSPLEX; TCP/IP is not used for that low level
communication.
-
Too busy driving to stop for gas!
Is there any point to having XCF communications between LPARs or would
TCPIP do just as well for that?
You must have XCF for Parallel SYSPLEX; TCP/IP is not used for that low level
communication.
I guess RFC2549 would be no good either then ... ???
Shane ...
I guess RFC2549 would be no good either then ... ???
Since I don't know what that is, I cannot answer that.
But, I know I would not want to use TCP/IP to handle GRS, DB2 buffer pools,
lock structures, etc.
-
Too busy driving to stop for gas!
-Original Message-
From: IBM Mainframe Discussion List
[mailto:[EMAIL PROTECTED] On Behalf Of Ted MacNEIL
Sent: Monday, April 21, 2008 8:53 AM
To: IBM-MAIN@BAMA.UA.EDU
Subject: Re: How fast is XCF
I guess RFC2549 would be no good either then ... ???
Since I don't know what
[EMAIL PROTECTED] (Shane) writes:
I guess RFC2549 would be no good either then ... ???
one of the april 1st RFCs
from my rfc index
http://www.garlic.com/~lynn/rfcietff.htm
click on Term (term-RFC#) in RFCs listed by section and scroll
down to April1
April1
5242 5241 4824 4042 4041 3751 3514
My company has a product that runs in muliple address spaces on multiple
LPARS and on windows unix boxes. Due to the relative response times of
XMS and TCPIP we've had to put the high msg activity ASIDs on on the same
LPAR as TCPIP is many orders of magnitude slower than XMS. We had hoped
What kind of path topology are you using for XCF? CTCs or ICF CF
structures? Or what?
And to what degree have you tuned XCF?
I'm not making any speed claim or promise. Just trying to understand your
situation in a little more detail.
Thanks, Martin
Martin Packer
Performance Consultant
IBM
Ron MacRae wrote:
My company has a product that runs in muliple address spaces on multiple
LPARS and on windows unix boxes. Due to the relative response times of
XMS and TCPIP we've had to put the high msg activity ASIDs on on the same
LPAR as TCPIP is many orders of magnitude slower than
13 matches
Mail list logo