Re: XRC Device Blocking - Where is it displayed in RMF?
FAO Anyone who's interested Just been told by IBM that XRC device blocking can be displayed in different places on different vendor hardware. IE HDS show device blocking in DISC time IBM ESS show it in IOSQ time. If using write pacing, then it's in DISC time for both HDS and IBM. Regards, JJ -- 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
Re: XRC Device Blocking - Where is it displayed in RMF?
Many thanks for that Derrick, We too have started looking at write pacing as we too are a multi vendor shop. We have found that HDS boxes only add 0.1 second for device blocking whereas our sharks add a whole second. We still haven't implemented write pacing so I guess that means our blocking delays are described in IOSQ. That seems to confirm my findings when looking at RMF. Now, how to break it gently to our performance team!! JJ -- 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
Re: XRC Device Blocking - Where is it displayed in RMF?
>From our experience with XRC, if we have performance problems on the host or primary DASD, it generally shows up as elongated IOSQ time (VERY elongated). Part of the issue is, if too much time is spent in a 'device blocking' state for a particular primary volume, WLM cant jump in an assign PAVs (we use dynamic PAVs) as quickly as they are needed. We are going to evaluate an alternative XRC pacing mechanism called 'write pacing' (recommended by IBM) that paces the host DASD devices by introducing small delays at the IO (channel program) level, rather than taking the volume offline for seconds at a time (which is the result of device blocking). IBM indicated that once 'write pacing' has been implemented, its possible to experience somewhat longer DISConnect times on primary DASD, but basically 'write pacing' is recommended as the superior XRC pacing mechanism. NOTE - not sure all OEM vendors support 'write pacing', we are inquiring with our vendors as we are a mixed shop. We have one sysplex with two members that are still using GRS ring setup, on these platforms, since we don't convert the SYSVTOC reserve, sometimes these usually short length reserves (should only be 1-2ms) end up taking several seconds on devices experiencing 'device blocking'. We intend to implement GRS Star to avoid this issue (convert the reserves). -- 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
Re: XRC Device Blocking - Where is it displayed in RMF?
Bill, My understanding (could be completely wrong, which is why I'm asking the question) is XRC will block a primary volume for a certain amount of time (the time the device is blocked is dependent on the vendor) if XRC can't keep up with the records that need to be written to the secondary volume. If the residuals go above a certain threshold (x'500' at our site)then XRC blocks any I/O to primary volume (unless specified as DONOTBLOCK). It has never been explained to me exactly how this happens. If I could find it in a manual (believe me, i've tried), I'd be a happy Storage Administrator. Bear in mind that XRC is Asynchronous. I can see that PPRC and synchronous copy delays are recorded in DISC time, but it says nothing about Asynchronous delays. The CMR is the Command Response from the CU to the IOS to say that it was ready for the I/O? I understood that until the CMR is recieved, the SSCH command is not issued. If the device blocking stops the CU from issueing a CMR, then IOSQ time would be where the delays are recorded as the SSCH hasn't been issued. Please feel free to pull my understanding to pieces as it's only through fully understanding what is happening that I can get to grips with the process and find out where our problems lay (lie?). Our performance team are saying that XRC device blocking is causing the problems as can be seen by the high DISC times. I don't believe that explanation (but am willing to eat humble pie if they're right and I'm wrong). JJ -- 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
Re: XRC Device Blocking - Where is it displayed in RMF?
In a message dated 11/10/2006 9:42:09 A.M. Central Standard Time, [EMAIL PROTECTED] writes: >I'm currently in a huge argument with the performance team who insist that >XRC device blocking is shown in DISC time. IOSQ time sounds more plausible >as the CU wouldn't reply to a CMR if the device was being actively blocked >so a SSCH wouldn't be issued? Disconnect time can only be accumulated by an I/O that has been started with SSCH. If an I/O is started, is blocked, and accumulates disconnect time, any other I/O request in the queue will accumulate IOSQ time. When such a queued I/O is started, it may or may not be blocked. I don't understand what XRC's device blocking is/means. Also what is CMR? But if for some reason IOS is not starting the first I/O request in the queue, then that request is certainly accumulating IOSQ time. There are many reasons why the next request might not be immediately started; e.g., UCB level is higher than normal and the next request's level is normal. 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
Re: XRC Device Blocking - Where is it displayed in RMF?
Paolo, Many thanks again. I'm currently in a huge argument with the performance team who insist that XRC device blocking is shown in DISC time. IOSQ time sounds more plausible as the CU wouldn't reply to a CMR if the device was being actively blocked so a SSCH wouldn't be issued? Is this documented anywhere? Regards, JJ -- 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
Re: XRC Device Blocking - Where is it displayed in RMF?
Joe,
XRC device blocking delays are included in IOSQ time.
_
Paolo Cacciari
Business Continuity and Recovery Services, IBM Global Services - South
Region, EMEA
Via Darwin 85, 20019 Settimo Milanese(MI) – Italy - MISET001
"The goal is to be prepared for a disaster not to continually plan for a
successful test"
* [EMAIL PROTECTED]
( + 39 051 41.36799 Mobile: + 39 335 6287584
7 + 39 02 596.23288
Joe jeffries
<[EMAIL PROTECTED]
COM> To
Sent by: IBM IBM-MAIN@BAMA.UA.EDU
Mainframe cc
Discussion List
<[EMAIL PROTECTED] Subject
.EDU> XRC Device Blocking - Where is it
displayed in RMF?
10/11/2006 14.14
Please respond to
IBM Mainframe
Discussion List
<[EMAIL PROTECTED]
.EDU>
Hi folks,
Wondered if anyone could give a difinitive answer to where RMF would show
delays caused by XRC device blocking.
Would it be in the DISC time, or the PEND time. Or maybe somewhere else
altogether.
Much appreciated.
JJ
--
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
--
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
XRC Device Blocking - Where is it displayed in RMF?
Hi folks, Wondered if anyone could give a difinitive answer to where RMF would show delays caused by XRC device blocking. Would it be in the DISC time, or the PEND time. Or maybe somewhere else altogether. Much appreciated. JJ -- 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
