Global Contention & false contentions

Nick Nur

Global Contention & false contentions
Hi Colleagues

How can I reduce real global contention in datasharing group where the
environment requires high level of concurrency?
I know only a binding parameter release (commit) in DB2 z/OS v8.
Is there any thing else I can do to reduce the amount of real lock global
contention?

thanks

Nick Nur

IMPORTANT NOTICE:

IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org

---------------------------------------------------------------------------------
Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm

Robert Catterall

Re: Global Contention & false contentions
(in response to Nick Nur)
Hey, Nick.

Real contention in a DB2 data sharing environment is just that -  the real deal.  It happens when, for example, some process wants to update a row in a page on which another process holds a lock.  In my mind, reducing real global lock contention in a DB2 for z/OS data sharing environment is similar to reducing contention in a non-data sharing DB2 system, a key difference being that you want to be more careful with the use of row-level locking in a data sharing group than you would in a stand-alone DB2 for z/OS environment, owing to the increased CPU overhead associated with row-level locking in a data sharing group (you can use row-level locking with DB2 data sharing -  just don't use it where you don't need it).

Do you currently have a problem with real global lock contention in your DB2 data sharing system?

Robert


Nick Nur wrote: Hi Colleagues How can I reduce real global contention in datasharing group where the environment requires high level of concurrency? I know only a binding parameter release (commit) in DB2 z/OS v8. Is there any thing else I can do to reduce the amount of real lock global contention? thanks Nick Nur IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm
--
Robert Catterall
President
Catterall Consulting
770-263-5322

www.catterallconsulting.com

IMPORTANT NOTICE:

IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org

--------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm

Nick Nur

Re: Global Contention & false contentions
(in response to Robert Catterall)
Thanks Robert.
Yes I see about 15% of real global contention and I do not know if that is bad or not. I like to reduce it for management sake to about 5% and I do not know how.
thanks again.


Robert Catterall <[login to unmask email]> wrote:
Hey, Nick.

Real contention in a DB2 data sharing environment is just that - the real deal. It happens when, for example, some process wants to update a row in a page on which another process holds a lock. In my mind, reducing real global lock contention in a DB2 for z/OS data sharing environment is similar to reducing contention in a non-data sharing DB2 system, a key difference being that you want to be more careful with the use of row-level locking in a data sharing group than you would in a stand-alone DB2 for z/OS environment, owing to the increased CPU overhead associated with row-level locking in a data sharing group (you can use row-level locking with DB2 data sharing - just don't use it where you don't need it).

Do you currently have a problem with real global lock contention in your DB2 data sharing system?

Robert


Nick Nur wrote:
Hi Colleagues How can I reduce real global contention in datasharing group where the environment requires high level of concurrency? I know only a binding parameter release (commit) in DB2 z/OS v8. Is there any thing else I can do to reduce the amount of real lock global contention? thanks Nick Nur IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the
latest on IDUG conferences at http://conferences.idug.org/index.cfm

--
Robert Catterall
President
Catterall Consulting
770-263-5322

www.catterallconsulting.com

IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm



---------------------------------
Looking for the perfect gift? Give the gift of Flickr!

IMPORTANT NOTICE:

IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org

---------------------------------------------------------------------------------
Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm

Robert Catterall

Re: Global Contention & false contentions
(in response to Nick Nur)
15% is not good.  Generally speaking, you want to see total global lock contention (real + false) at 5% or less.  Are you seeing a high number of DB2 for z/OS deadlocks and or timeouts?  Looking at a DB2 monitor accounting detail report, or an online display of thread detail information, what's the wait for global lock time (one of the so-called class 3 wait time categories) as a percentage of in-DB2 (also known as class 2) elapsed time?

Robert


Nicola Nur wrote:
Thanks Robert.
Yes I see about 15% of real global contention and I do not know if that is bad or not. I like to reduce it for management sake to about 5% and I do not know how.
thanks again.


Robert Catterall <[login to unmask email]> wrote:
Hey, Nick.

Real contention in a DB2 data sharing environment is just that -  the real deal.  It happens when, for example, some process wants to update a row in a page on which another process holds a lock.  In my mind, reducing real global lock contention in a DB2 for z/OS data sharing environment is similar to reducing contention in a non-data sharing DB2 system, a key difference being that you want to be more careful with the use of row-level locking in a data sharing group than you would in a stand-alone DB2 for z/OS environment, owing to the increased CPU overhead associated with row-level locking in a data sharing group (you can use row-level locking with DB2 data sharing -  just don't use it where you don't need it).

Do you currently have a problem with real global lock contention in your DB2 data sharing system?

Robert


Nick Nur wrote: Hi Colleagues How can I reduce real global contention in datasharing group where the environment requires high level of concurrency? I know only a binding parameter release (commit) in DB2 z/OS v8. Is there any thing else I can do to reduce the amount of real lock global contention? thanks Nick Nur IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm
--
Robert Catterall
President
Catterall Consulting
770-263-5322

www.catterallconsulting.com

IMPORTANT NOTICE:
IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org
--------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email]. Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm


Looking for the perfect gift? Give the gift of Flickr! IMPORTANT NOTICE:

IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org

--------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email]. Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm


--
Robert Catterall
President
Catterall Consulting
770-263-5322

www.catterallconsulting.com

IMPORTANT NOTICE:

IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org

--------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm

Nick Nur

Re: Global Contention & false contentions
(in response to Robert Catterall)
Thanks again Robert
I am doing accounting package long report and in some instances the TOTAL CL8 SUSPENS. is a large subset of the ELAP-CL7 TIME-AVG and a large component of TOTAL CL8 SUSPENS. is the GLOBAL CONTENTION.
I do not see lots of timeouts or lots of deadlocks. But we have an application that requires a high degree of concurrency and our system is a mixture of page and row level locking and there are lots of requests for locks in a very dynamic environment.
I think I got the idea now.I shall concentrate on reducing the the amount of lock requests as if it is a non datasharing environment then the global contention of the datasharing will be reduced. Thanks Robert again.


Robert Catterall <[login to unmask email]> wrote:
15% is not good. Generally speaking, you want to see total global lock contention (real + false) at 5% or less. Are you seeing a high number of DB2 for z/OS deadlocks and or timeouts? Looking at a DB2 monitor accounting detail report, or an online display of thread detail information, what's the wait for global lock time (one of the so-called class 3 wait time categories) as a percentage of in-DB2 (also known as class 2) elapsed time?

Robert


Nicola Nur wrote: Thanks Robert.
Yes I see about 15% of real global contention and I do not know if that is bad or not. I like to reduce it for management sake to about 5% and I do not know how.
thanks again.


Robert Catterall <[login to unmask email]> wrote:
Hey, Nick.

Real contention in a DB2 data sharing environment is just that - the real deal. It happens when, for example, some process wants to update a row in a page on which another process holds a lock. In my mind, reducing real global lock contention in a DB2 for z/OS data sharing environment is similar to reducing contention in a non-data sharing DB2 system, a key difference being that you want to be more careful with the use of row-level locking in a data sharing group than you would in a stand-alone DB2 for z/OS environment, owing to the increased CPU overhead associated with row-level locking in a data sharing group (you can use row-level locking with DB2 data sharing - just don't use it where you don't need it).

Do you currently have a problem with real global lock contention in your DB2 data sharing system?

Robert


Nick Nur wrote:
Hi Colleagues How can I reduce real global contention in datasharing group where the environment requires high level of concurrency? I know only a binding parameter release (commit) in DB2 z/OS v8. Is there any thing else I can do to reduce the amount of real lock global contention? thanks Nick Nur IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out
the latest on IDUG conferences at http://conferences.idug.org/index.cfm

--
Robert Catterall
President
Catterall Consulting
770-263-5322

www.catterallconsulting.com

IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm




---------------------------------
Looking for the perfect gift? Give the gift of Flickr! IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org
--------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm

--
Robert Catterall
President
Catterall Consulting
770-263-5322

www.catterallconsulting.com

IMPORTANT NOTICE: IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org --------------------------------------------------------------------------------- Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm



---------------------------------
Looking for the perfect gift? Give the gift of Flickr!

IMPORTANT NOTICE:

IDUG is pleased to announce a series of upgrades to the DB2-L discussion listserv that are being implemented to improve reliability and the overall user experience of DB2-L. These changes are coming on November 30th. Details at http://www.idug.org

---------------------------------------------------------------------------------
Welcome to the IDUG DB2-L list. To unsubscribe, go to the archives and home page at http://www.idugdb2-l.org/archives/db2-l.html. From that page select "Join or Leave the list". The IDUG DB2-L FAQ is at http://www.idugdb2-l.org. The IDUG List Admins can be reached at [login to unmask email] Find out the latest on IDUG conferences at http://conferences.idug.org/index.cfm