db2 ats (advanced technical support) -...

DB2 ATS (Advanced Technical Support)

© 2015 IBM Corporation

John Iczkovits

[email protected]

SWARUG

z/OS 2.1 and 2.2 Great New Features for DB2

September 30, 2015


© 2015 IBM Corporation2

Disclaimer

� © Copyright IBM Corporation 2015. All rights reserv ed. U.S. Government Users Restricted Rights - Use, duplication or disclosure r estricted by GSA ADP Schedule Contract with IBM Corp.

� THE INFORMATION CONTAINED IN THIS PRESENTATION IS P ROVIDED FOR INFORMATIONAL PURPOSES ONLY. WHILE EFFORTS WERE MAD E TO VERIFY THE COMPLETENESS AND ACCURACY OF THE INFORMATION CO NTAINED IN THIS PRESENTATION, IT IS PROVIDED “AS IS” WITHOUT WAR RANTY OF ANY KIND, EXPRESS OR IMPLIED. IN ADDITION, THIS INFORMA TION IS BASED ON IBM’S CURRENT PRODUCT PLANS AND STRATEGY, WHICH ARE SUBJECT TO CHANGE BY IBM WITHOUT NOTICE. IBM SHALL NOT BE RESP ONSIBLE FOR ANY DAMAGES ARISING OUT OF THE USE OF, OR OTHERWISE REL ATED TO, THIS PRESENTATION OR ANY OTHER DOCUMENTATION. NOTHING CO NTAINED IN THIS PRESENTATION IS INTENDED TO, NOR SHALL HAVE TH E EFFECT OF, CREATING ANY WARRANTIES OR REPRESENTATIONS FROM IBM (OR ITS SUPPLIERS OR LICENSORS), OR ALTERING THE TERMS AND CONDITIONS OF ANY AGREEMENT OR LICENSE GOVERNING THE USE OF IBM P RODUCTS AND/OR SOFTWARE.

� IBM, the IBM logo, ibm.com, DB2 are trademarks or r egistered trademarks of International Business Machines Corporation in the United States, other countries, or both. If these and other IBM trademarked terms a re marked on their first occurrence in this information with a trademark sym bol (® or ™), these symbols indicate U.S. registered or common law trademarks o wned by IBM at the time this information was published. Such trademarks may also be registered or common law trademarks in other countries. A current list o f IBM trademarks is available on the Web at “Copyright and trademark information” at www.ibm.com/legal/copytrade.shtml



dfp Enhancements and DB2

� As with DB2, dfp introduces enhancements on a period ic basis. These enhancements may effect data sets used by DB2 , however, keep in mind that many times a DB2 APAR is required before DB2 can take advantage of the dfp enhancement. – For example, the dfp enhancement to increase the VSAM data set maximum

to 7,257 extents required some DB2 APARs in order to utilize the enhancement. Another example is when dfp introduced large sequential data sets in z/OS 1.7, DB2 could not take advantage of utilizing this enhancement for such things as archive log data sets and image copies until the introduction of DB2 9.

– This is true for other dfp enhancements as well. For example, extended format sequential in z/OS 1.11 is not available for use for utilities with the current versions of DB2 (8 and 9).

� Verify if any DB2 maintenance is required to take a dvantage of dfpenhancements.

� DB2 generally releases a new version every 3 years. MVS (z) on the other hand provides a release about every two y ears. There are 1 or several disk related updates once a year. Make sure you understand how these other newer releases relate to DB2.



4

z/OS® DFSMStm V2.1 Highlights (GA September 30, 2013)

� DFSMShsm– DFSMS Storage Tiers– DFSMShsm Tape Enhancements– DFSMShsm Fast Replication

Enhancements– DFSMShsm RAS & Usability

Enhancements

� Catalog– RLS Support for Catalog– Catalog Contention Detection

Enhancements– Catalog Alias Enhancements– Catalog CSI Enhancements– Catalog Parmlib Member

Enhancements– Catalog DFSMS GDG

Enhancements– Catalog: RNLs HealthCheck

� IDCAMS– Larger BlockSize for LBI Support– IDCAMS Support for RLS– DELETE PDS/PDSE with Mask– ALTER NULLIFY Management

Class– DIAGNOSE of GDGs

� Access Methods– VSAM RLS Directory Only

Caching– VSAM RLS 64-bit Enhancements– VSAM RLS Dynamic Volume

Count– VSAM SMB Enhancements– VSAM SHOWCB Enhancements– DFSMS Support for zHPF– SAM EF Support for FlashCopy

� PDSE– PDSE Version 2 Enhancements– PDSE Larger Member Size– GDG support for PDSEs– IEBCOPY Enhancements

� SMS – Provide Accurate Volume Space

Statistics– Alter ACDS/COMMDS to

SHAREOPTIONS(3 3)– New PARMLIB Option to

Generate PDSE– SMS ACS Read-only Variable for

EAVs



z/OS® DFSMStm V2.1 Highlights

� DFSMSdfp– OCE Partial Release

Enhancements– OCE RAS Enhancements– XTIOT HealthCheck

� DSS– Reset with RESTORE– zFS Change Activity Support

� SDM– XRC Offline Volumes

� OAM– OAM Usability & Reliability

Enhancements

� DFSORT– Dynamic Sort Enhancements– Functional Enhancements– 64-bit Enhancements



Messages Value z/OS ® V2.1 Enhancements Which Companies Benefit?

Data and Analytics

Ready

Customers can manage significantly more data for lessCustomers can process active content at scale

� Compression zEDC� SMC-R (RoCE) � VSAM RLS enhancements� Batch enhancements � Fonts

� Companies who need to securely process huge volumes of data.

� Companies who need to analyze more active data, move data efficiently and quickly.

Cloud Ready

Drive higher throughput, remarkable availability, and scale like no other system

� 2GB pages� Flash Express � Transactional Execution� 100 way SMP� RMF enhancements

� Companies with advanced, or complex workloads requiring extreme availability resiliency and scale.

� Companies who need to connect private clouds with back end systems and data

Security Ready Secure your infrastructure, protect assets and data, leverage crypto as a serviceAudit and comply with regulations.

� Crypto as a service� New health checks� PKCS #11 and EMV standards� Improved auditing, compliance� Designed for EAL 5+

� Service Providers, Telcos, Banks, or any company who needs highly secured data with centralized policy based management

� Companies that need to comply with regulations and improve audit readiness

Mobile Ready Extend mainframe application data securely to mobile devices

� DB2/CICS® JSON interfaces� Worklight® for delivery to devices� Workload isolation� Secured , high speed

communications

� Customers with systems of record requiring mobile access to secured content and enforced isolation of workloads

6

z/OS V2.1 quick reference



IBM continues to redefine and extend the role of the mainframe to deliver new strategic capabilities and deeper client value

7

Cloud ReadyHighly scalable, agileheterogeneous enterpriseprivate cloud

� Elastic, virtually limitless expansion� Shared everything design� Secure hybrid computing with

centralized management� Built on open industry standards

Data ReadyEnterprise data repository that integrates operational analytics for accelerated insight� Enterprise data hub� High-volume secure and reliable

transaction processing� Integrated, real-time operational

analytics

Mobile ReadyConnecting backend systems to mobile devices to turn each interaction into an opportunity

� Build and connect to back end systems� Secure and manage with the highest

level of security� Extend and transform capabilities to

mobile devices

Security ReadyTrusted security and reliability for critical business processes,applications and data

� Ultimate data security protection� Built-in cryptography� Unmatched resiliency and

availability� Support for industry standards



Notes:1 The IBM Lifecycle Extension for z/OS provides the a bility for customers

to purchase extended defect support for that releas e of z/OS for up to 24 months after the z/OS release’s end of service d ate

2 See IBM GTS services for additional fee-based exten ded service3 Optional extended service is planned to be offered• Planned. All statements regarding IBM's plans, dire ctions, and intent

are subject to change or withdrawal without notice. WdfM – Server has been withdrawn from Marketing

Release

z900/z800

WdfM

z990/ z890

WdfM

z9 EC

z9 BC

WdfM

z10 EC

z10 BC

WdfMz196 CPC

z196 w/zB

Xz114 CPC

z114 w/zB

X

zEC12

zBC12 z13

End of Servic

e

Coexist

with

Lifecycle Extension for

z/OS

z/OS V1.10 X X X X X X X X X 9/111 R12 9/131

z/OS V1.11 X X X X X X X X X 9/12* R13 9/141

z/OS V1.12 X X X X X X X X X X 9/14* 2.1 3*z/OS V1.13 X X X X X X X X X X 9/16* 2.2 3*z/OS V2.1 X X X X X X X X 9/18* 2.3 3*z/OS V2.2* x x X X X X X 9/20* 2.4* 3*

z/OS Support Summary

Out of Lifecycle Extension for z/OS support

Defect support provided with Lifecycle Extension for z/OS

Generally supported

* Planned. All statements regarding IBM's plans, dire ctions, and intent are subject to change or withdrawal without notice.



IDCAMS – Management Class – ALTER NULLIFY

� ALTER NULLIFY Management Class

– Currently, IDCAMS ALTER does not allow a user to nullify a management class, but ISMF provides support to remove the management class from a dataset by specifying '-'.

– New enhancement: Allow a user to specify 'ALTER NULLIFY' to nullify a Management Class. • Users can specify NULLIFY(MANAGEMENTCLASS) to

nullify the Management Class of a dataset. • The abbreviation is NULLIFY(MGMTCLAS).• Can alter a data set’s Management Class to null

�Why it Matters: Improved usability of the ALTER function to match similar function in ISMF



10

IDCAMS - LBI

� Larger BlockSize for LBI Support

– The large block interface (LBI) was introduced ten years ago.• It uses less tape storage and transfers data faster

– IDCAMS has not supported LBI, and this restriction can stop the use of the LBI feature• IDCAMS REPRO and PRINT utilities fails with error message

IDC3300I followed by IDC3321I and a return code of 12.• The max block size for IDCAMS REPRO and PRINT is 32760.

– New enhancement: IDCAMS REPRO and PRINT will support a block size up to the access method limit, which currently is 256 KB.• The block size is still limited to 32 KB when the data set does not

support LBI, such as with a unit record device or TSO terminal.• SYSIN and SYSPRINT do not support LBI.



Improving Usability and Skills

� GDGs in chronological order!

• New GDGORDER JCL DD statement keyword• Can specify that generations be returned from oldest to newest• No need to sort or concatenate!• System default remains newest-to-oldest• Why it Matters: Allows all the members of a GDG to be processed in

chronological order without being sorted (Default is LIFO, now FIFO is possible). May reduce some DB2 jobs time based on order.

� ISPF potpourri (a partial list of enhancements):

• Edit support for Unicode data• Edit support for an expandable command field• Edit HILITE command to highlight the invalid lowercase JCL characters• Edit support for regular expressions in FIND and CHANGE commands• Support for dynamically allocated data sets using XTIOTs for EDIT,

BROWSE, LMINIT, and LIBDEF• Improved enhanced member list function• ISPF directory list display for z/OS UNIX, UDLIST, DIRLIST support for a

SRCHFOR function• Support for multiple logical screens on ISPF entry, and multi-screen exit

when ending ISPF• Path name mask support in the z/OS UNIX Directory List Utility• Support in OPT3.4 for a “free” line command for multivolume data sets• Support in UDLIST lower-case path names



12

IDCAMS - GDGs� DIAGNOSE of GDGs

– Currently, IDCAMS DIAGNOSE command does not detect a mismatch in the actual number of extension cells versus how many GDG BCS records exist. • This can lead to GDG processing errors that are

undetectable until batch processing fails.

– New enhancement: Enhance AMS DIAGNOSE to crosscheck the extension cells and the GDG BCS record, so the mismatch can be detected.• DIAGNOSE will return a failing return code, if there is a

mismatch of the actual number of extension cells and the number of BCS records of a GDG.

�Why it Matters: Intended to help users easily and quickly identify the cause of GDG processing errors.



13

IDCAMS� DELETE PDS/PDSE with Mask

– In R12, IDCAMS provided a DELETE option to delete all members of a partitioned data set in a single operation.

– New enhancement: Enhance the IDCAMS DELETE command to be more flexible in performing the deletion of the members in a partitioned data set (PDS/PDSE).

• Allows a mask for member names to be specified in a DELETE command for PDS/PDSEs.

• A mask for a member name can contain an asterisk (*) or percent sign (%).– Asterisk means 0 or more characters– % means 1 and only 1 character

• Double asterisk (**) still means delete all members in the PDS/PDSE.• Examples:

─DELETE SOME.DATA.SET(EELLS*) – …to delete all members starting with “EELLS”─DELETE SOME.DATA.SET(EELLS%A)

– …to delete all members with EELLSxA, where x is any character



14

PDSEIEBCOPY Enhancements

– Currently, IEBCOPY can be used to copy to or from a PDSE data set a member and its aliases, together as a group. Copy functions require that member names be fully qualified.

• COPYGRP supports all combinations of group copy requests for PDSE to PDSE, PDSE to PDS and PDS to PDSE, but PDS to PDS group copies are currently treated as a COPY operation.

– New enhancement: New COPYGROUP function designed to copy members and all aliases for any combination of PDS and PDSE data sets.

• A superset of the existing COPYGRP function.• All aliases in a group will be copied with the member or neither

the aliases nor the member in a group will be copied.• The EXCLUDE statement is not supported.

– New enhancement: Allow the user to pass a filter pattern mask on COPYGROUP functions

• The MEMBER sub parameter of the SELECT and EXCLUDE statements will be enhanced to accept the “*” and '%' filter control characters.

�Why it Matters: Improved usability for IEBCOPY.



15

PDSE

� PDSE Larger Member Size– Currently PDSE members are limited to 15,728,639 records.

• Message IEC036I 002-A8 is issued if a PDSE member exceeds 15,728,639 records. This limit does not exist for PDS datasets.

– New enhancement: Increase the limit on PDSE member size.

• PDSE member size is planned to be over 125 times larger (approximately 2,146,435,071 records) than the current limit in many circumstances, and substantially larger than the maximum supported size of a PDS member.

• Be sure to look at toleration PTFs

– SHARE Requirement met: SSMVSS11010

�Why it Matters: Provide additional scalability and usability benefits of using PDSEs in place of PDSs and make it feasible to use PDSEs instead of multiple large sequential data sets.



16

SMS - PDSE� New PARMLIB Option to Generate PDSE

– Currently, the DSNTYPE={LIB|HFS} does not guarantee that a partitioned data set will be created

• A partitioned data set is created if DSORG=PO or directory blocks are specified

• Otherwise, a sequential data set is created.– New enhancement: SMS provides a new parameter in IGDSMSxx that

directs SMS to create a PDSE:

• HONOR_DSNTYPE_PDSE(YES|NO); NO is the default and the processing remains as before

• HONOR_DSNTYPE_PDSE(YES) applies only when DSNTYPE={LIB|HFS} – A PDSE is created regardless of the specification of DSORG and directory blocks when

HONOR_DSNTYPE_PDSE(YES)

• SMS provides a new SETSMS command to modify the setting

�Why it Matters: Allows the user to specify that partitioned data sets be unconditionally allocated as PDSE when DSNTYPE=LIBRARY is specified, whether or not directory space is also specified in JCL.



PDSE V2

� PDSE Member-Level Recovery• Keep multiple generations of each PDSE member• Recover a prior version using ISPF edit or the DESERV API• SMS DATACLAS, DESERV API support• Specify maximum member generations at PDSE Version 2 allocation time• System-wide maximum specification with a new MAXGENS_LIMIT keyword in IGDSMSxx� The support for PDSE member generations is provided in z/OS V2R1 as a small

programming enhancement (SPE) via APAR OA42358. The ISPF support is provided with APAR OA42248.

�PDSE Version 2• Designed to improve read performance, reduce storage consumption

• New PDSE member size limit over 125 times larger in most cases, and substantially larger than the maximum size of a PDS member

• Intended to make it possible provide additional scalability and usability benefits of using PDSEs in place of PDSs and make it feasible to use PDSEs instead of multiple large sequential data sets

• ISPF using DSLIST can be used to maneuver between versions of a PDSE V2 data set, but IBM Data Set Commander (DSC) for z/OS, V8.1 (formerly known as IBM ISPF Productivity Tool for z/OS) provides extensive support for version 2 PDSE data sets with member generations.

• Keep in mind, libraries such as SDSNSAMP are PDS, not PDSE. You cannot use PDSE V2 as a way of keeping multiple versions of ZPARMs for example. Technical limitations include not being able to use TSO EDIT for PDSEs.



18

PDSE� PDSE Version 2 Enhancements

– New enhancement: Simplify PDSEs to be able to make partial release more effective and to improve general PDSE performance.• The system will create a new PDSE format,

Format 2, which will allow all unused space to be released, consolidate directory pages when possible, improve read performance, and reduce virtual storage utilization for PDSE processing.– Toleration of the new PDSE format in z/OS V1.12 and z/OS V1.13

• Overall PDSE performance will be improved. – The path length of almost all PDSE related

operations will be reduced, and index searches will be improved.

– Unnecessary structures from the directory will be removed allowing space to be used more efficiently.



Batch Modernization…• Dynamic ENQ downgrade support in GRS, and JCL support:─ In a multistep job, change an exclusive ENQ to shared ENQ for a data set

– After the last job step with DISP=OLD, MOD, or NEW has ended─ New JES2 Job Class parameter, DSENQSHR=AUTO|ALLOW|DISALLOW─ New JOB statement parameter, DSENQSHR=ALLOW, to use with ALLOW

//GREAT JOB (accounting), DSENQSHR=ALLOW

//STEP1 EXEC PGM=WHATEVER

//OLD DD DSN=MY.DATA.SET, DISP=NEW,…

//STEP2 EXEC PGM=SOMEPGM

//STILLOLD DD DSN=MY.DATA.SET, DISP=MOD

//STEP3 EXEC PGM=EXPCT806

//SHR4NOW DD DSN=MY.DATA.SET, DISP=SHR

//STEP4 EXEC PGM=IDUNNO

//OLDAGAIN DD DSN=MY.DATA.SET, DISP=OLD

//STEP5 EXEC PGM=NOCLUE

//SHR4EVER DD DSN=MY.DATA.SET, DISP=SHR

//STEP6 EXEC PGM=WHOKNOWS

//STILLSHR DD DSN=MY.DATA.SET, DISP=SHR

Exclusive ENQ until last DISP=OLD, NEW, or MOD step done

Then, shared ENQ



Even more Batch Modernization…

• JES2 symbols support for instream data and Submit

─New step-level EXPORT statement to list system and JCL symbols available to be resolved, and new callable service support for access to them

─New SYMBOLS keyword for DD * and DD DATA to control substitution

Example:

// EXPORT SYMLIST=(DSNAME)

// SET DSNAME=MY.DATA.SET

// SET VOLSER=VOLUME

//*

//DELETEDS EXEC PGM=IDCAMS,REGION=300K,

//SYSPRINT DD SYSOUT=*

//DEVICE DD DSN= &DSNAME,VOLUME=&VOLUME,DISP=OLD

//SYSIN DD *, SYMBOLS=JCL

DELETE –

&DSNAME. -

NONVSAM -

PURGE -

SCRATCH -

FILE(DEVICE)

/*



Still more Batch Modernization…

• New PARMDD EXEC keyword support for longer parameter strings

─ Mutually exclusive with PARM keyword─ No other changes required for unauthorized programs─ Authorized programs must be bound using LONGPARM or system will terminate the

job at step initiation─ Supports parameter lists from 1 to 32,760 bytes long

Example://NOTAREAL JOB (accounting info),MSGLEVEL=(1,1),CLASS=BATCHLOW,

// NOTIFY=&SYSUID

//*

//UNAUTH EXEC PGM=MYPGM,PARMDD=PARMS

//IN DD DISP=SHR,DSN=MY.DATA.SET

//OUT DD DISP=(,CATLOG),DSN=MY.NEW.DATA.SET, …

//PRINT DD SYSOUT=*

// PARMS DD *

LONG PARAMETER LIST HERE IN THE DATA SET NAMED BY PARMDD.

NOTE THAT IT NEED NOT BE AN INSTREAM DATA SET. A SEQUENTIAL

DATA SET, A MEMBER OF A PDS OR PDSE, OR Z/OS UNIX FILE WILL

WORK AS WELL. AND, IF I COUNTED RIGHT, THEN THIS VERY VERY

LONG PARAMETER LIST IS NOW WELL OVER 100 CHARACTERS IN LENGTH

AND I CAN STOP TYPING!

/*



Allocation updated to enable you to specify that migrated data sets be allocated by a batch job be recalled in parallel , before each job step starts

─ Issue recall requests during step initiation

─Wait for all recalls to complete

─Continue with Allocation processing needed to start the stepNew

ALLOCxx Parmlib: BATCH_RCLMIGDS(PARALLEL)SETALLOC: SETALLOC

SYSTEM,BATCH_RCLMIGDS=PARALLEL

Yet more Batch Modernization !!

//STEP xyz

// DD DSNAME= BATCH.DS1

// DD DSNAME=BATCH.DS2




Recall 30sec

Recall 30sec

Recall 30sec

Recall 30sec

Recall 30sec

//STEP xyz

// DD DSNAME= BATCH.DS1





Recall 30sec

Recall 30sec

Recall 30sec

Recall 30sec

Recall 30sec

Elapsed Recall Time = 150 seconds Elapsed Recall Tim e = 30 seconds



Improving Usability and Skills

�Catalog alias processing improvements:• Data sets with (NONVSAM) aliases defined using

SYMBOLICRELATE to be searched for in the catalog “owning” the high-level qualifier

• Creation dates to be stored in alias entries and listed by IDCAMS

• Catalog connector alias entries to be kept when you temporarily delete a user catalog

� “ Improved IEF212I message”• Really, it’s a new message:• IEFA107I JOBNAME PROCNAME STEPNAME DDNAME -

DATA SET NO.SUCH.DATA.SET NOT FOUND• (Instead of IEF212I … DDNAME + 009)



24

� The new Storage Group Processing Priority specifies the relative order in which storage groups should b e processed during Automatic Space Management and Automatic Backup Processing

– Storage Groups will be processed in the order of their priority

• A higher value means a higher priority• Valid values are 1-100• Default value is 50

80 50 50 20

SMS Storage Group Priority



RLS for ICF Catalogs housing DB2 data sets

� At times, DB2 performance is degraded due to access of the ICF catalog

� Record-level Sharing for the ICF catalog provides:

– Most ICF catalog contention likely to evaporate

– Expected to remove most size and performance-related reasons for splitting ICF catalogs in a Parallel Sysplex

� RLS is a complete replacement for other ICF catalog performance features such as the Enhanced Catalog Sharing(ECS) or VLF/IS C caching

– Master catalog is not RLS-eligible

� CF is required for RLS even for a stand alone non-D ata Sharing subsystem

� RLS requires that the ICF catalog be SMS managed

� Great potential to increase performance for DB2 dat a sets, especially when many data sets are accessed concurrently

� Should help with batch job performance as well as R EORGs during switch phase when many data sets are involved.



26

IDCAMS� IDCAMS Support for RLS

– Customer requirement to copy, print and backup VSAM data sets while sharing the VSAM data set with other applications.

– New enhancement: Enhance IDCAMS commands PRINT, REPRO, IMPORT and EXPORT to be able to open VSAM data sets using RLS.• A new optional keyword RLSSOURCE

({NO|YES| QUIESCE}) and/or RLSTARGET ({NO|YES|QUIESCE}) will be implemented for the PRINT, REPRO, IMPORT and EXPORT commands.

– SHARE Requirement Met: SSMVSS01007



CA RECLAIM � Enhance VSAM and VSAM RLS to reclaim empty

space (control areas) in KSDSs. Applies to SMS and non-SMS data sets.

– Will resolve the main reasons for KSDS reorganizations:

– Reclaim space

– Improve both sequential and direct performance

– Can be used to reclaim space in an ICF catalog if running out of space or extents without down time

� Introduced with z/OS 1.12

� For KSDS data sets only. Recommendation is for the ICF catalogs housing DB2 data sets to have a Data Class specification of CA RECLAIM turned on.



zEnterprise Data Compression (zEDC) - can help to reduce CPU and storageEvery day 2.5 quintillion bytes of data are created

2828

* The amount of data sent to an SMF logstream can be reduced by up to 75% using zEDC compression – reducing logger overhead** These results are based on projections and measurements completed in a controlled environment. Results may vary by customer based on individual

workload, configuration and software levels*** All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.

z/OS V2.1 zEDC

zEDC Express

Efficiently compress active data by providing a low CPU, high performance, dedicated compression accelerator.

Industry standard compression for cross platform data distribution **

Typical Client Use Cases:Significant disk savings with trivial CPU cost for large

BSAM/QSAM sequential files More efficiently store audit data in application logs

Reduce the amount of data needed for data migration and backup/restore **

Transparent acceleration of Java compressed applications **

Compress your data

4X*(efficient system data

compression)

Up to 118X reduction

in CPU and up to 24X throughput improvement when zlib uses zEDC **

Data Ready

z/VM 6.3 support for guests***



IBM z Enterprise Data Compression New data compression offering that can reduce resource usage

What Changes?It is time to revisit your decisions about compression.

What is it? � zEDC Express is an IO

adapter that does high performance industry standard compression

� Used by z/OS Operating System components, IBM

Middleware and ISV products

� Applications can use zEDCvia industry standard APIs (zlib and Java)

� Each zEDC Express sharable across 15 LPARs,

up to 8 devices per CEC.

� Raw throughput up to 1 GB/s per zEDC Express Hardware Adapter

� QSAM/BSAM compression can save up to 4x disk space and in some cases shorten elapsed time, reducing batch windows.

� HSM/DSS Offload host-based compression

� Business Partner Data Exchange can have higher throughput with lower CPU cost

� Managed File Transfer saves up to 4x link bandwidth, and up to 80%

elapsed time.

� ISV Products delivery expanded customer value

� Java for z/OS R7.1 accelerates common compression classes used by applications and middleware

� Improved availability with SMF compression.

� Disk Savings : Many people are already getting value from CMPSC compression and software compression today

� Performance : High throughput alternative to existing System Z compression for large or active files.

� Industry Standard: Low cost compressed data exchange across all platforms

� Pervasive: Standard APIs allow quick adoption by middleware products running on System Z

What is the Value?New sources of customer value



Three Ways to Compress (and Decompress) on z/OS

�Software compression– CPU-intensive

– Much slower

– Data can be inflated on anything supporting the same algorithm

�Compression coprocessor-based instructions– Dictionary-based compression, generic or tailored

– Can be inflated on a System z processor

– All compression consumes apparent CP cycles

• Compression done on the coprocessor, but accounted for as CP busy time because the CP is unavailable until the coprocessor is done

�New zEDC Express adapter for zEC12 and zBC12 and zEnterprise Data Compression (zEDC) for z/OS V2.1– Compression work is offloaded to the card

– Minimal CP cycles consumed

– zlib-based, industry-standard deflate compression

– Data can be inflated anywhere zlib processing is available



zEnterprise Data Compression

� Now available:− zEDC Card & z/OS feature− zBNA support− SMF and RMF support− Support for industry standard zlib

compression− zlib library in z/OS V2.1− SMF data compression on z/OS V2.1− Software-based decompression

support for SMF data on z/OS V1.12 and V1.13

− Java support − IBM Encryption Facility support − Extended Format BSAM/QSAM support− WebSphere MQSeries support− IBM Sterling Connect:Direct support



SMF Data Compression

�For SMF data written to log streams– We expect about a 4:1 compression ratio for SMF data

– Designed to significantly increase SMF recording rates

– Can specify that all SMF data or SMF data written to selected log streams be compressed

– New SMFPRMxx COMPRESS keyword on LSNAME and DEFAULTLSNAME

– New PERMFIX subparameter of COMPRESS to balance fix/unfix overhead with available real memory

�Corresponding IFASMFDL support– Automatic inflation on z/OS V2.1 with feature and Hardware

support

– SOFTINFLATE parameter for software-based decompression

─For z/OS V1.12 & z/OS V1.13, with the PTF for APAR OA41156

─ Included in z/OS V2.1─ Intended to be used when zEDC is not available



IBM System z Batch Network Analyzer and CompressionHelping determine if you have files that are candidates for zEDC

� IBM System z Batch Network Analyizer 1.4.6

– A free, “as is” tool to analyze batch window

– PC based, and provides graphical and text reports

• Including Gantt charts and support for Alternate Processors

� Available from IBM Washington Systems Center

– IBMers:http://w3.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS5126

– BP’s:https://ibm.com/partnerworld/wps/servlet/mem/ContentHandler/tech_PRS5133

– Customers:http://ibm.com/support/techdocs/atsmastr.nsf/WebIndex/PRS5132

� zBNA identifies zEDC Compression Candidates

– Post-process customer provided SMF records, to identify jobs and data sets which are zEDCcompression candidates across a specified time window, typically a batch window

– Help estimate utilization of a zEDC feature and help estimate the number of features needed

– Generate a list of data sets by job which already do hardware compression and may be candidates for zEDC

– Generate a list of data sets by job which may be zEDC candidates but are not in extended format



More Compression Support

�Extended Format BSAM and QSAM Compression– New support for Compressed Format data sets now available with the PTF for

APAR OA42195

– In addition to generic (DBBLIB) and tailored (supply a dictionary) compression

– New COMPACTION option in DATACLAS definition

– New values on COMPRESS parameter in IGDSMSxx

� DFSMSdss ™ data compression – DUMP, COPY, and when DFSMSdss is used as the data mover by DFSMShsm™ see

APAR OA42243

– When a disk output data set is used

* Statements regarding IBM future direction and intent are subject to change or withdrawal, and represent goals and objectives only.

* Statements regarding IBM future direction and intent are subject to change or withdrawal, and represent goals and objectives only.



zEDC – what it means to DB2� Tables and indexes do not use zEDC compression

� Archive log data sets can be compressed

– Performance could be a problem if the archive logs are being read in any way other than forward sequential

– Archive log data sets can also be compressed using the SMS Data Class attribute COMPACT

� SMF

– Can still use ZPARM SMFCOMP for standard compression

– zEDC not used for records running OMPE/PM (IBM’s Omegamon for DB2)

� BSAM/QSAM - IBM DB2 utilities and tools do not use zEDC for compression

– DB2 image copies can take advantage of standard DB2 compression



36

z/OSMF V2.1: smarter software management at a glanceImproved visibility Into software currency with new Software Management

� Visualize end of service information

to help improve software management� Check SMP/E inventory information

against catalog entries and other data sets

� Provide reporting functions to plan software maintenance

� Easily identify end of service across the enterprise

� See all your products across your enterprise at a glance

� Readily pinpoint whether maintenance, fixes are installed or needed

� Gain visibility into installed software– Understand your software currency– Plan ahead to create improved lifecycle support– Make it easier to create “gold” instances for standardize deployments– Manage software smarter--reduce costs, improving quality

With z/OSMF manage software currency checking across the enterprise

in less than 15 minutes versus days



z/OS Management Facility Can Help� IBM z/OS Management Facility (z/OSMF) delivers on

the IBM’s strategy for mainframe simplification and modernization

� Is a free feature of z/OS

� Provides a modern browser based interface for managing the z/OS system.

� z/OSMF workflows can be customized once, executed multiple times

� Extended to simplify management of z/OS subsystems like DB2

� z/OSMF workflows can help DB2 staff to

– Migration multiple DB2 subsystems easily

– Deploy a new DB2 subsystem quickly



Coming Soon – zOSMF Default DB2 Migration Workflow

� DB2 will generate the default workflow artifacts through an updated DB 2 migration CLIST

– Workflow definition file

– Set of new JCL templates as the z/OSMF artifacts used for DB2 11 migration

– Workflow variable file

� DB2 users can edit the sample workflow input variable file with the input for a specific DB 2 subsystem

� Workflow can be customized by adding/removing/changing steps

� zOSMF to support REST calls over workflows to enable cloud-based solutions such as automated test/dev configuration 38



39

Flash Express improves system availability and performance during key transition periods

19%Reduction in total

dump time for 36 GB standalone dump

10xFaster response time and

37% increase in throughput compared to disk for morning

transition

28%Improvement in DB2®

throughput leveraging Flash Express with Pageable Large

Pages (PLP)

Flash Express:Improve availability and performance during workload

transition and spikes

Faster, less disruptive diagnostics

Cost effective, resilient solution for overflow of MQSeries®

shared queues in Coupling Facility (SOD) *

Supported on z/OS V1.13 or Higher**

Dynamic Reconfiguration

for Flash Express

Flash Express feature for zEC12

and zBC12

All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.** With the March 2013 RSM Enablement Offering Web deliverable

Up to



Flash Express Support�Available for z/OS V1.13 with…

• A zEC12, zBC12, z13 server with Flash Express

• z/OS V1R13 RSM Enablement Offering web deliverable

─http://www.ibm.com/systems/z/os/zos/downloads/

• Dynamic Reconfiguration and optional PLPA/COMMON page data sets in enabling PTFs

• All these functions are included in z/OS V2.1

�z/OS to use Flash Express for:• Pageable large pages

• Paging, when performance would be improved vs. disk-based paging

• SVC and StandAlone Dump

• CF cache structure for WebSphere MQ



Large (1MB) Page Support

�To use 1MB pages, you need…• An IBM System z10® or later server

• z/OS R12 or later, to use fixed large pages

• z/OS V1R13 RSM Enablement Offering web deliverable to use pageable large pages on zEC12 and zBC12 servers

�Current exploiters of fixed large pages:• Java™ 6 SR1 and later, and its exploiters

• Including WebSphere Application Server

• z/OS R12 and later XL C/C++ programs using Language Environment®

• The z/OS operating system, in z/OS R12 and up

• IBM DB2® 10 for z/OS (5605-DB2) and up

�Exploiters for pageable large pages:• z/OS V1.13 and z/OS V2.1 Language Environment (with a runtime option)

• A maintenance roll-up of IBM 31-bit and 64-bit SDK7 for z/OS Java Technology Edition, Version 7 (5655-W43 and 5655-W44)

• DB2 10 and DB2 11 with the PTF for APAR PM85944

• IMS™ 12 (5635-A03) Common Queue Server, with the PTF for APAR PM66866

� 1MB pages are first come first serve. Make sure you have enough pages for all of your requirements

� Starting with DB2 11 ZPARM OUTBUFF uses 1MB pages w hen available– There is no mechanism to turn this off

– Keep in mind, OUTBUFF is allocated in MSTR and is allocated before the buffer pools in DBM1



Large Page Frames

• Each DB2 subsystem/member “owns” its own page. For example, Subsystem DB2P BP10 may own a large page that no other subsystem/member or buffer pool will share.

• When deemed inefficient, some pages will be demoted to a smaller frame size. For example, BP10 has 10,000 4K pages and is now 1MB. See next page.

• In this case, instead of all 10,000 pages fitting into 1MB pages, 16 –4K pages were allocated to avoid wasting memory.

• Carefully configure buffer pools to avoid waste of memory. • MVS command D VIRTSTOR,LFAREA is for an entire LPAR. See

next page for more information. • Efficiency can be determined by reviewing SMF 113 records and

review using RMF “CPU Measurement Facility” (CPU MF).



-DB2P DISPLAY BUFFERPOOL(BP4) SERVICE(4) DSNB401I -DB2P BUFFERPOOL NAME BP4, BUFFERPOOL ID 4, USE COUNT 245 DSNB402I -DB2P BUFFER POOL SIZE = 10000 BUFFERS AUTOSIZE = NO 895

ALLOCATED = 10000 TO BE DELETED = 0 IN-USE/UPDATED = 369

DSNB406I -DB2P PGFIX ATTRIBUTE - 896 CURRENT = YES PENDING = YES

PAGE STEALING METHOD = LRU DSNB404I -DB2P THRESHOLDS - 897

VP SEQUENTIAL = 80 DEFERRED WRITE = 50 VERTICAL DEFERRED WRT =

10, 0 PARALLEL SEQUENTIAL =50 ASSISTING PARALLEL

SEQT= 0 DSNB999I -DB2P DSNB1DBP SERVICE( 4 )OUTPUT DSNB999I -DB2P 4K PAGES 16 DSNB999I -DB2P 1M PAGES 9984 DSN9022I -DB2P DSNB1CMD '-DISPLAY BUFFERPOOL' NORMAL COMPLETION

D VIRTSTOR,LFAREA IAR019I 08.51.07 DISPLAY VIRTSTOR 904 SOURCE = 06 TOTAL LFAREA = 6144M , 0G LFAREA AVAILABLE = 2656M , 0G LFAREA ALLOCATED (1M) = 3488M LFAREA ALLOCATED (4K) = 0M MAX LFAREA ALLOCATED (1M) = 3632M MAX LFAREA ALLOCATED (4K) = 0M LFAREA ALLOCATED (PAGEABLE1M) = 0M MAX LFAREA ALLOCATED (PAGEABLE1M) = 0M

SERVICE(4) may not be required for DB2 11.

Note that 16 – 4K pages were allocated to avoid wasting an entire 1MB page.

Command is for the entire LPAR

4K pages denoted are for when there is not enough memory, NOT that 4K pages were used.



IBM zEC12/zBC12 and z13�2GB fixed page frames

• If 1 MB pages are good…• …sometimes 2 GB pages are even better!• Exploited by IBM 31-bit SDK for z/OS,

Java Technology Edition, V7.0.0 (5655-W43) and SDK IBM 64-bit SDK for z/OS, Java Technology Edition, V7.0.0 (5655-W44)

• Used by DB2 11 for buffer pools• Available for other large structures, other

users• Supported on z/OS V2.1 or on z/OS

V1.13 with the RSM enablement web deliverable and the PTF for APAR OA40967

� In order to verify usefulness of pages sizes, collect SMF type 113 records and analyze using CPU MF - CPU Measurement Facility



Buffer Pool enhancements …� Buffer pool FRAMESIZE attribute

– DB2 11 allows FRAMESIZE(2G)

• PGFIX(YES) required• Requires z/OS 2.1 & EC12, BC12, z13 (possible retro to 1.13) with OA40967 • If PGFIX(NO), then buffer pool allocated using 4K pages

– DB2 11 allows FRAMESIZE(1M) with PGFIX(NO)

• Control blocks pageable (PMB – Page Manipulation Blocks)

– LFAREA to include control blocks– ((VPSIZE*pagesize/1024/1024) + 6399)/6400 = # of MBs of

storage used for the PMBs• Data pages will remain 4K• Requires EC12, BC12, Z13 and only a CEC that is Fla sh Express capable

in order to be able to exploit pageable large frames – otherwise demoted to 4K. Note – FE is not required, only the capabilit y!

• CPU savings by reducing Translation Lookaside Buffer access

– When you cannot perform PGFIX(YES)



Buffer Pool enhancements… Frame size Page fix Supported DB2 H/W Requirement B enefit

4K NO All N/A Most flexible configuration

4K YES All N/A CPU reduction during I/O

1M NO DB2 10 with APAR, or

DB2 11

zEC12 and Flash Express capable

Backed by real or LFAREA

CPU reduction from TLB hit

Control Blocks only, also applies to 2G Framesize

1M YES DB2 10 above z10 above

LFAREA 1M=xx

CPU reduction during I/O, CPU reduction from TLB hit

2G YES DB2 11 zEC12

LFAREA 2G=xx

CPU reduction during I/O, CPU reduction from TLB hit

* If any HW/SW requirements unmet, 4K frames used



Group Buffer Pool Write Around Protocol ...

� Circumvent situations where significant, sustained page write workload causes

– Application slow downs

– Pages to be placed in the LPL in extreme cases

� DB2 detects large write activity and

– Write to the Group Bufferpool if the page is already cached

– Otherwise write are directed to DASD

• 50% of GBP or 20% of Class

� These are actions beyond the use of the Auto Alter option



Group Buffer Pool Write Around Protocol

� Normal GBP use is restored when the storage shortage is relieved

– 40% of GBP or 10% of Class

� Available in CM without a REBIND, with

– 1.13 with OA37550 PTFs or

– z/OS 2.1 and

– CFLEVEL=17

� DISPLAY GROUPBUFFERPOOL ... MDETAIL

– WRITE-AROUND PAGES instrumentation in DSNB777I

– Also in IFCIDs 0002 & 0003



Parallel Sysplex - Two New Functions� “ Thin” interrupts for CFs

• CFCC polled for work to do, so all CF engines were 100% used

• So, shared engine CFs had a very limited use case

─A CF always used its entire PR/SM™ timeslice except in DYNDISP=YES mode─Not a great way to implement many/most production CFs

• New CFCC design in CFLEVEL 19 on zEC12 and zBC12 servers along with XES/XCF changes designed to use a more interrupt-driven, hyrbrid approach

• Allow the use of shared engine coupling facilities in many environments with acceptable performance

• Intended to lower Parallel Sysplex entry costs by reducing the number of environments for which dedicated coupling facility (CF) engines are needed to achieve good performance

� Also, new set of “thin” interrupts to be used by z/OS• Designed to decrease response time on the average

• Help reduce XCF and XES processing overhead and improve performance when processing asynchronous coupling facility operations and recognizing certain CF events

� Available on z/OS V1.12 and V1.13 with the PTFs for APARs OA38734, OA38781, OA37186, & OA42682



�CF structure rebuild performance• Prior to z/OS V2.1, all CF structures were rebuilt in parallel

when duplexing is initiated• Considerable contention may result, slowing the process

overall and (especially) slowing the process for the most important structures

• New design to process structures serially, more or less• Intent is much faster recovery for critical structures and faster

overall rebuild time• System structures to be prioritized by the system• Other structures optionally prioritized by policy

Scalability and Performance



51

Optimize server to server networking – transparently“HiperSockets™- like” capability across systems

z/OS V2.1 SMC-R

10GbE RoCE Express

z/VM 6.3 support for guests*

zBC12zEC12

Up to 50% CPU savings for FTP file

transfers across z/OS systems versus

standard TCP/IP **

Up to 48% reduction in response time and

10% CPU savings for a sample CICS

workload exploiting IPIC using SMC-R versus

TCP/IP ***

Up to 40% reduction in overall transaction

response time for WAS workload accessing

z/OS DB2 **** * All statements regarding IBM's future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only.** Based on internal IBM benchmarks in a controlled environment using z/OS V2R1 Communications Server FTP client and FTP server, transferring a 1.2GB binary file using SMC-R (10GbE RoCE Express feature) vs standard TCP/IP (10GbE OSA Express4 feature). The actual CPU savings any user will experience may vary.*** Based on internal IBM benchmarks using a modeled CICS workload driving a CICS transaction that performs 5 DPL (Distributed Program Link) calls to a CICS region on a remote z/OS system via CICS IP interconnectivity (IPIC), using 32K input/output containers. Response times and CPU savings measured on z/OS system initiating the DPL calls. The actual response times and CPU savings any user will experience will vary.**** Based on projections and measurements completed in a controlled environment. Results may vary by customer based on individual workload, configuration and software levels.

Shared Memory Communications (SMC-R):Exploit RDMA over Converged Ethernet (RoCE) with qualities of

service support for dynamic failover to redundant hardware

Typical Client Use Cases:Help to reduce both latency and CPU resource consumption over

traditional TCP/IP for communications across z/OS systemsAny z/OS TCP sockets based workload can seamlessly use

SMC-R without requiring any application changes



RDMA over Converged Ethernet (RoCE)

�RoCE Support for SMC-R • Requires z/OS V2.1 running on zEC12,

zBC12, z13 servers with the RoCE Express feature

• Shares memory between peer z/OS images

• Read/write access to the same memory buffers without application changes

• Designed to help increase transaction rates with low latency and reduced CPU cost

• RMF support with new SMF74-9 records and PCIE Activity Report

• Java support in IBM 31-bit and 64-bit SDK for z/OS Java Technology Edition, Version 7 (Java7R1, 5655-W43 and 5655-W44)



� WLM improvements:

• New types of classification groups and qualifier types you can use to define rules like SPM more consistently

• Support for up to 3,000 application environments, an increase from the prior limit of 999

� RMF enhancements:

• RMF designed to offload some processing to zIIP processors in a Parallel Sysplex (when a zIIP is available)

• 1 MB page and Flash Express reporting enhancements ─ Also available on z/OS V1.13 with the RSM Enablement Offering web deliverable and

the PTF for APAR OA38660

• Support for new interrupt delay time measurement on zEC12 and zBC12 systems─ SMF74-1 and SMF79-9 support─ z/OS V1.12 and V1.13 support with PTF for APAR OA39993

• Global Mirror collision reporting in RMF Monitor I and SMF74-5 records─ z/OS V1.12 and V1.13 support with PTF for APAR OA40376

• More information about CF links in Monitor I─ z/OS V1.12 and V1.13 support with PTF for APAR OA37826

Self-Managing Capabilities



Exploit Enhanced Client Information Fields …

� Longer client information fields exploited for dist ributed applications

– Pre V11 - value set by the API is truncated

– V11 - value set by the API supports extended lengths (NFM)

– Tolerated in CM, still truncated to16 bytes (same as pre V11)

– Longer client information fields are exploited in NFM + APPLCOMPAT V11R1

� Longer field support for special registers:

– CLIENT_USERID supports up to 128 bytes

– CLIENT_WRKSTNNAME supports up to 255 bytes

– CLIENT_APPLNAME supports up to 255 bytes

– CLIENT_ACCTNG supports up to 255 bytes

� Longer fields supported in DDF traces, messages and displays in CM



Exploit Enhanced Client Information Fields …

� Resource Limit Facility (RLF) can govern SQL statement activity based on longer client informati on fields in NFM

� System Profile Monitoring supports longer client information fields in NFM

� WLM classification rules can utilize longer client information fields when running z/OS V 2R1



56

DFSORT� Dynamic Sort Enhancements

– While DFSORT has installation defaults designed to control the use of central storage resources by all SORT applications running concurrently on a system, customers still often experience degraded performance due to over commitment of resources and high paging. • This is usually due to changes in available resources after a sort has already committed to using large amounts of

central storage.

– Often the first submitted DFSORT job can possibly use most of the central storage for creating dataspaces, hiperspaces or memory objects. • Any DFSORT job(s), submitted later, use very little or no dataspace, hiperspace or memory objects because the

central storage is not available at that time.

– New enhancement: A new TUNE option is designed to allow users to specify that DFSORT obtain storage incrementally and check on storage availability before allocating additional storage.• Helps to better balance utilization for sort operations and other workloads initiated within a short time.

– New enhancement: Increase the memory object work space maximum from 64 GB to 1 TB, to allow users to sort larger amounts of data in memory object work files.

�Why it Matters: Improved memory resource management to better balance the memory requirements of multiple large concurrent sort operations and other workloads.



57

DFSORT� Functional Enhancements– Alphanumeric Tests

• New enhancement: Support new alphanumeric tests, using binary format, for both compare fields and parse fields, including combinations of alphanumeric character sets (uppercase and lowercase, and numeric).

�Why it Matters: Allows users to specify various sets of characters using a single compare condition or PARSE keyword rather than using compare conditions or PARSE keywords.

– Symbol and PARSE Enhancements

• New enhancement: Allow symbols to be used for more DFSORT operands, and the number of parse fields supported is planned to be increased from the prior limit of 100 fields to 1,000 fields.

�Why it Matters: Improved DFSORT usability

– Add String at End of Variable-Length Records

• New enhancement: Allow users to specify that a string up to 50 characters in length be appended to variable-length output records.

�Why it Matters: Improved DFSORT usability.



DB2 11 - RACF Synchronization of Security Caches� When

– z/OS 1.13 with OA39486, OA39487, OA39506

– In CM

– DSNX@XAC is active

– AUTHEXIT_CACHEREFRESH=ALL (NONE is the default)

• Not online updateable

– ENF 79 requires Resource Class SIGNAL=YES in RACF Class Descriptor Table

• When not using IBM supplied resource classes for DB2 objects & Admin Authorities

• YES is default for IBM supplied RACF resources• Single and Multi-system scope class names supported

� RACF sends notifications to DB2 about these changes



59

� Today: No support for renaming of a VSAM data set d uring recovery from dump.

� V2R1: Add Support

– Existing optional parameter NEWNAME(newdsname) can now be specified for VSAM data sets when recovering with FROMDUMP

– DFSMSdss RENAMEU and REPLACEU enhancements for this support

• Physical data set COPY and RESTORE will be enhanced to support renaming of VSAM data sets when RENAMEUnconditional is specified. Renaming of AIX will remain unsupported.

• Physical data set RESTORE will be enhanced to support REPLACEUnconditional.

– Caveat – Data set can only be cataloged under certain conditions

dss recoveryRecover VSAM data set with a NEWNAME



zHyperWritefor z/OS, DB2 and DS8870� New zHyperWritefunction for DB2, z/OS and DS8870 wit h GDPS or TPC-

R HyperSwap

– Delivered year end 2014

– Leverages synergy of z/OS and DS8870 replication technologies

– Designed to accelerate DB2 Log Writes in Metro Mirror environment

� •Benefits include:

– Improved DB2 transactional latency

– Log throughput improvement

– Additional headroom for growth

– Improved resilience for workload spikes

– Potential cost savings from workload consolidation

� DB2 commit response time reduced up to 40%

– Benefit percentage varies with distance

� •Requires:

– zHyperWritefunction in z/OS 2.1, with the PTF for APAR OA45662

– DB2 10 and DB2 11 with PTF for APAR PI25747

– IBM DS8870 Storage Subsystem MCL60



61

The classic DFSMS storage hierarchy (35 years old)

High Performance & Function

LowCost

$$$Level 0

(L0)

$$Migration Level 1

(ML1)

$Migration Level 2

(ML2)

Data in PrimaryLevel is Managed

by DFSMShsm

Data in MigrationLevels is “Owned”

By DFSMShsm

Storage Tiers: V2R1Background



� The Space Management Environment is evolving

� Typical configurations have changed to leave data on Level 0 longer and then migrate directly to ML2

� When ML2 is a VTS, the VTS disk cache replaces the ML1 tier

Eliminates MIPS required for software compression to ML1Eliminates DFSMShsm ML1->ML2 processing

� Tape storage controller functionality replacing DFSMShsm duplexing

� Shortcomings of today's DFSMS functionality

�No policy-based automation for moving data within the Primary Storage Hierarchy (Level 0)

�No policy-based management of Active (open) data

Storage Tiers: V2R1Background



z/OS V2R1 DFSMS provides the Storage Tiers Solution. ..

Automated, policy-based space management that moves data from tier to tier within the Primary (Level 0) Hierarchy

�Automated movement provided via the existing DFSMShsmSpace Management function

� Movement is referred to as a 'Class Transition‘

� Data remains in its original format and can be immediately accessed after the movement is complete

�Policies implemented via the existing Class Transition policies and updated Management Class policies

�Enhanced support for DB2, CICS and zFS data

– - Open data temporarily closed to enable movement

Storage Tiers: V2R1Overview



Primary Storage Hierarchy

Migration Hierarchy

ML2(VTS)

Smart Tier 0:

SSD / Enterprise

($x)

Allocate

180 Day Migration

Recall

Hot

Warm

Luke-warm

Cool

Cold

Frigid

Data “Temperature”

45 Day Transition

Smart Tier 1:

Enterprise / Nearline($1/3x)

This example environment shows data being “transitio ned” from ‘Smart Tier 0’ to ‘Smart Tier 1’ after 45 days and then becoming eligible for migration after 180 days of inactivity.

Tier 0: ‘Hot’ data is moved to SSD, but ‘Cold’ data is never allowed below Enterprise Class storage.

Tier 1: ‘Hot’ data is not allowed higher than Enterprise Class, and ‘Cold’ data is allowed to reside on Nearline.

Storage Tiers: V2R1Overview



65

Access Methods� SAM EF Support for FlashCopy

– Multi-volume, single-striped extended format sequential data set cannot use FlashCopy with DFSMSdss because the volume boundary cannot be changed.• The suffix in each block contains the block number on that volume, and the

FlashCopy process cannot adjust those numbers.

– Similarly a single-volume, single-striped extended format data set cannot use FlashCopy if the destination requires multiple volumes.

– New enhancement: Support the use of FlashCopy by removing volume boundary awareness from sequential extended format data sets. • The system will create a new extended format, Format 2, of an extended format data set which

contains no volume boundary awareness within the suffix associated with each physical block. • The data set's catalog entry will indicate the data set's version and LISTCAT and DCOLLECT will

display the version.• Do not set Extended Format Version 2 until all sha ring and backup systems are at z/OS

V2.1 or compatibility PTFs have been installed.

�Why it Matters: Removes a long standing restriction and allows the exploitation of FlashCopy with single striped extended format data sets.



FRRECOV DS to Any Volume: z/OS V2R1

� Today

– FRRECOV DSNAME requires that a data set be recovered to the volume(s) that the data resided on at the time of the backup

� Enhancement

– If there is not enough space on a volume to perform the recovery, HSM will select the volume in the same Storage Group that has the most free space



FRRECOV DS to NEWNAME: z/OS V2R1� Today

– A data set must be recovered using the same name

� Enhancement

– A data set may be recovered to a new name

• Recovery may from either Disk or Tape

FRRECOV DSNAME(ds.oldname) NEWNAME(ds.newname)

FRRECOV DSNAME(ds.oldname) NEWNAME(ds.newname) REPLACE

– If there isn’t enough space on the original volumes, then new volumes will be selected



SDM� XRC (ie z/OS Global Mirror) Offline Volumes

– The current implementation of XRC requires that primary volumes (the volumes being written to by application programs) be online to the XRC system when the volumes are added to an XRC session, and when a XRC session is restarted.

• As customer installations have grown from hundreds, to thousands, to tens of thousands of primary volumes, the time to vary these volumes online has become quite large. – “Vary online processing before XSTART takes excessive amount of time, and if vary

'missed', XADD processing fails.”

– “Volumes offline, XDELPAIR fails after XSTART, unable to locate UCBs”

– New enhancement: Allow XRC primary volumes to be offline when the XSTART and XADDPAIR commands are issued to start or restart mirroring for existing volumes.

�Why it Matters: improve availability by eliminating the need to wait for all devices to be varied online.



Consistency Groups: z/OS V2R1

• Standard FlashCopy – ‘Fuzzy’ Backup• DB2 ensures data consistency during recovery

A A’

B B’

C C’

D D’

• FC SrcA to TrgA’• Writes held to SrcA during establish• Write activity on B, C, D

• FC SrcB to TrgB’• Writes held to SrcB during establish• Write activity on A, C, D

• FC SrcC to TrgC’• Writes held to SrcC during establish• Write activity on A, B, D

• FC SrcD to TrgD’• Writes held to SrcD during establish• Write activity on A, B, C




� Freeze FlashCopy Consistency Group

– I/O activity to a volume in a consistency group is held in an extended long busy (ELB)

– A window can be created during which dependent write updates will not occur

– I/O activity resumes when a FlashCopy Consistency Group is created

� Consistency Group support is intended for DB2 Logs

– Default ELB is two minutes

– Create a consistent copy of the Log copy pool such that a conditional restart is not required




• Consistency Group FlashCopy – Consistent copy

A A’

B B’

C C’

D D’

• FC SrcA to TrgA’ / CGFREEZE• Writes held to SrcA by ELB• Non Dependent write activity on B, C, D

• FC SrcB to TrgB’• Writes held to SrcA, B by ELB• Non Dependent write activity on C, D

• FC SrcC to TrgC’• Writes held to SrcA, B, C by ELB• Non Dependent write activity on D

• FC SrcD to TrgD’• TrgA – D contain a consistent copy

• Consistency Group Created / CGTHAW



Consistency Groups: z/OS V2R1� Gotchas

– The Consistency Group THAW is an LSS level operation

• The control unit does not enable logical volume connected Freezes

• If two copy pools with volumes in the same LSS are being backed up concurrently, the copy pool to first issue the THAW will invalidate the backup of the other copy pool

• FRBACKUP requests are single threaded on an HSM host, so the exposure is when FRBACKUP requests are issued from multiple HSM hosts

• HSM recognizes the issue and will fail the backup, which will have to be reissued

– If the time to establish the FlashCopies for the volumes being processed as a part of the consistency group exceeds the CG timer (2 min default), then the control unit will release the ELBs on all of the volumes and the backup will fail




Panel Utilities Scroll Help

------------------------------------------------------------------------------ DGTDCPP1 COPY POOL DEFINE/ALTER Page 1 of 5

Command ===>

SCDS Name . . . : USER6.MYSCDS

Copy Pool Name : COPYPL1

To DEFINE/ALTER Copy Pool, Specify:

Description ==> ___________________________________ _________________________

==> ____________________________________________________________

Auto Dump . . . N (Y or N) Dump Sys/Sys Group Name . . . ________

Dump Class . . ________ Dump Class . . ________

Dump Class . . ________ Dump Class . . ________

Dump Class . . ________

Number of DASD Fast Replication Backup

Versions with Background Copy . . . . . . . . 2 (0 to 85 or blank)

FRBACKUP to PPRC Primary Volumes allowed . . . __ ( NO, PN, PP, PR or blank)

FRRECOV to PPRC Primary Volumes allowed . . . __ (N O, PN, PP, PR or blank)

FlashCopy Consistency Group . . . . . . . . . N (Y o r N)

Use ENTER to Perform Verification; Use DOWN Command to View next Panel;

Use HELP Command for Help; Use END Command to Save and Exit; CANCEL to Exit.

• LIST COPYPOOL will have an indicator of the setting



DB2 and SMT2

� Requires z/OS 2.1

� Requires z13

� Concern - slower engine speeds on the DBM1 SRBs may affect prefetch – must verify!

– For such things as DRDA transactions, this means that you potentially have more transactions completing in a given period, but each transaction may run slower.

� For performance, check IIPCP time in RMF

77



z/OS® DFSMStm V2.2 GA Highlights (September 2015)

� DFSMShsm– DFSMS Storage Tiers

Enhancements– Fast Replication Enhancements– DFSMShsm / DFSMSdss zEDC

Exploitation

� Catalog / IDCAMS– Support for > 255 GDGs– MODIFY CATALOG Security

Enhancement– Catalog RAS Enhancements– IDCAMS VERFIY RECOVER– IDCAMS RAS Enhancements

� SMS– Space Constraint Relief– RAS Enhancements– Storage Group Space Alert

Messages

� Access Methods– RLS Index Level Locking– RLS Large Memory Exploitation– Secondary Space Reduction– BSAM/QSAM zEDC Exploitation

78

� Open/Close/EOV– Dynamic Exits– RAS Enhancements

� DADSM/CVAF– RAS Enhancements

� DFSORT – zHPF Exploitation– Functional Enhancements

� DFSMSrmm– WHILECATALOG Support– SEARCHDATASET /

SEARCHVOLUME Enhancements

� DFSMS / Storage Synergy– Multi-Target PPRC– IBM zHyperWritetm

– Easy Tier® Application Hints– XRC Workload Write Pacing– Multiple Incremental FlashCopy



DFSMS V2.2 Statements of Direction

� DFSMSrmm– z/OS V2.2 is planned to be the last release to provide

support in the Common Information Model (CIM) component for the Javatm Managed Provider Interface (JPMI).

– z/OS V2.2 is planned ot be the last release to support the DFSMSrmm CIM Provider.

� System Data Mover (SDM)– z/OS V2.2 is planned to be the last release to include a

number of TSO/E-based SDM related commands. • Does NOT effect query commands (CQUERY, FCQUERY,

RQUERY, XQUERY, XSTATUS) or the XSET command.– IBM recommends you use the REXX version of theses

commands (introduced with z/OS R12).– DB2 customers using FlashCopy or System Level Backup

(SLB) should consider changing their TSO/E commands to REXX if not using commands such as ZQUERY.

79



DFSMShsm V2R2

DFSMShsm Fast Replication Enhancements– Dump (Command, Auto, Fast Replication) commands today must be

processed on the same DFSMShsm host that initiated the request. • The resources of a single DFSMShsm host do not scale to the large copy pool

sizes that need to be processed.

– New enhancement: Distribute dump work across all hosts in a sysplex-wide Common Dump Queue (CDQ).• Introduces an architecture to distribute workload across host members in

an HSMPlex, and the new architecture allows for flexible configurations.• Invoked and controlled via SETSYS commands.

�Why it Matters: Improves overall throughput by distributing the workload across the sysplex instead of concentrating it on a single LPAR.

• Copying pools can now be accomplished much faster when hsm is configured across different LPARs in the SYSPLEX using CDQ.

80



Dump StackingDEFINE DUMPCLASS

� Added a new MINSTACK, minimum stack, parameter and an MAXSTACK alias for the STACK parameter

|___ _______________________ ___ ________________ ___||__ _STACK____ _(maxv)__| |__MINSTACK(minv)_|

|_MAXSTACK_|

� maxv represents the preferred maximum number of dump cop ies that the system should use to place on a tape volume

� minv represents the preferred minimum number of dump cop ies that the system should use to place on a tape volume– Utilizes more dump tasks, but may have less dump copies on tape

� MAXSTACK value must be equal to or larger than the MINSTACK value

� If MINSTACK value not specified the MAXSTACK value used for itExample, DEFINE DUMPCLASS(DCLASS1 MAXSTACK(100) MINSTACK(20))

81



Dump Stacking

82

STACK and MINSTACK Use Case

–400 volumes to be dumped

–Dump class defined with MAXSTACK(100)

–5 dump tasks

–Today

• 100 volumes are stacked onto 4 tapes, utilizing 4 dump tasks

–Enhancement

• Add MINSTACK(20) to dump class

• Volumes dumped in sets of 20 across all available dump tasks (5)

• Additional volumes added to each tape in sets of 20 until MAX reached

–80 volumes will be stacked onto 5 tapes, utilizing all 5 dump tasks

■ MINSTACK may also be specified on the BACKVOL command



DFSMShsm

� DFSMS Storage Tiers Enhancements

– Today, the Class Transition function is limited to automatic space management processing.

– New enhancement : Enable Class Transitions to be initiated with the various Migration commands.

• The various Migrate commands are enhanced to support class transitions at the data set, volume and storage group level.

– New keywords: BOTH, MIGRATIONONLY, TRANSITIONONLY

– The default behavior is to perform both migration and transition processing for VOLUME and STORAGEGROUP operations.

– The default for MIGRATE DATASET is to perform a migration. The TRANSITION keyword indicates that a transition should be performed.

�Why it Matters: Storage Tiers in general allows users to better align storage costs with changing business value and minimize the TCO of System z data. The new enhancements extend the use of tiering and improve the usability of the function.

• DB2 customers have a more flexible way of moving data.

83



DFSMShsm� DFSMS Additional Storage Tiers Enhancements

– Today, for “classic” migration, command initiated volume migration is single threaded. This severely limits the throughput.

– New enhancement : New STORAGEGROUP keyword for the MIGRATE command to initiate migration for all volumes within a storage group in parallel.• MIGRATE STORAGEGROUP(sgname, sgname, …) …

• Up to 30 storage group may be specified

• Primary Space Management processing will be performed for the storage groups, as opposed to the standard volume-by-volume processing performed by MIGRATE VOLUME.

– Use Case : Clients create DB2 image copies (using FlashCopy) to a specific storage group and then invoke DFSMShsm MIGRATE to move the copies to tape.

– As another example, you can now migrate your DB2 archive log Storage Group with one command.

– This enhancement enables a single command to process all of the volumes in the storage group in parallel .

�Why it Matters: Significantly improve the throughput for command initiated space management and improve usability.

84



DFSMShsm

� DFSMS Additional Storage Tiers Enhancements– Today, moving data to newly defined disk volumes within a storage

environment can be manually intensive and cumbersome.

– New enhancement : Enable the DFSMS Class Transition function to move data laterally to the same tier of storage, in addition to the ability of moving data to different tiers storage.

• MOVE keyword added to the MIGRATE DSNAME, VOLUME and STORAGEGROUP commands.

– Every data set will be processed, regardless of management class policy or threshold, and ACS routines will be invoked to determine the new storage class and/or storage group.

– Use Case: Move DB2 data from existing smaller volumes to the new larger, newly defined EAVs.

�Why it Matters: Simplify the task of migrating data to newly defined disk volumes. 85



DFSMShsm

� DFSMShsm Fast Replication Dump Enhancements– Today, updating the expiration date is manually intensive, as it

can take thousands of FIXCDS commands to update the expiration date of a single copy pool dump version.

– New enhancements:• Provide a new command to update the expiration date for a copy pool

dump copy.– New UPDTCDS command updates ALL of the CDS records that need to be

updated to reflect the new expiration date.

– Avoid using hundreds of FIXCDS commands.

�Why it Matters: Optimize Dump throughput and improve usability. 8

6



Dump Expiration

87

■ New UPDTCDS command will update the expiration date for a copy pool dump copy

–Today, it can take thousands of FIXCDS commands to update the expiration date of a single copy pool dump version

UPDTCDS COPYPOOL(name ) VERSION(ver )DUMPEXPIRATION(DCLASS( name ) NEWDATE(date ))

–This new command updates all of the CDS records that need to be updated to reflect the new expiration date.

• The Fast Replication Dump record

• The dump record for each volume dumped

–This command will be extended to other functions in the future



DFSMShsm

� DFSMShsm Fast Replication Messaging Enhancements– Today, finding messages related to a Fast Replication command is very difficult

because they are written to shared logs.

• First, find the correct log and then search through all of the messages for all HSM activity.

– New enhancement: Provide an option to dynamically allocate a unique message data set and write all DFSMShsm and DFSMSdss messages for a specific FRBACKUP or copy pool Automatic Dump to that data set.• Invoked via SETSYS command

– SETSYS FASTREPLICATION(MESSAGEDATASET(YES|NO HLQ(hl q)))

– Indicates that all Fast Replication messages should go to a unique data set with a unique HLQ.

– Each message is prefixed with the date and timestamp, along with the ID of the host the wrote the message.

– Message Data Sets must be SMS managed.

– SHARE Requirement Addressed: SSMVSS14005

�Why it Matters: Greatly simplify problem determination.

• NOTE! Not yet automatically available for System Level Backup, but can be used with native FlashCopy. .

88



DFSMShsm

� DFSMShsm/DFSMSdss Exploitation of zEDC– New enhancements: DSS and HSM now exploit zEDC for dumping and

restoring data and when HSM uses DSS to move data.• DSS

– Supports DUMP, RESTORE, COPY, DEFRAG, CONSOLIDATE, RELEASE, and PRINT of zEDC compressed format data sets

– Supports new compression keyword for DUMP> Accepted for all FULL, TRACKS, physical and logical DATASET backups to DASD and tape

• HSM– Supports MIGRATION, RECALL, BACKUP, RECOVER, FULL-VOLUME DUMP,

RECOVER FROMDUMP, FRBACKUP, FRRECOV, ABACKUP, ARECOVER of zEDC compressed format data sets

– Uses the DFSMSdss zEDC support for certain functions.

�Why it Matters: Provides efficient compression and lower CPU overhead than the processor-based and software-based compression methods already available.*

89

* Disclaimer: Based on projections and/or measurements completed in a controlled environment. Results may vary by customer based on individual workload, configuration and software levels.



90

ML1 zEDC Compression – Value!

* Disclaimer: Based on projections and/or measurements completed in a controlled environment. Results may vary by c ustomer based on individual workload, configuration and software levels.

HSM Compaction z EDC0

10

20

30

40

0

1

2

3

4

5

HSM Migrate to ML1 DASD

CPU (s)

Elapsed (s)

Size (GB)

Tim

e (

S)

Ou

tpu

t Siz

e (

GB

)

Up to 80% CPU ReductionUp to 50% Space Reduction

½ the data for SSM½ the data to Recycle

90



HSM Compact zEDC0

20

40

60

80

100

120

HSM Recall from ML1 DASD

CPU (s)

Elapsed (s)

Tim

e (

S)

91

ML1 zEDC Compression – Value!

* Disclaimer: Based on projections and/or measurements completed in a controlled environment. Results may vary by c ustomer based on individual workload, configuration and sof tware levels.

Up to 65% CPU Reduction

91



92

APAR OA48394 ML2 zEDC Compression

* Disclaimer: Based on projections and/or measurements completed in a controlled environment. Results may vary by c ustomer based on individual workload, configuration and sof tware levels.

CU Comp zEDC0

20

40

60

80

100

0

1

2

3

4

5

HSM Migrate to ML2 (TS7720 - HYDRA)

CPU (s)

Elapsed (s)

Hydra Size (GB)

Tim

e (S

)

Ou

tpu

t Siz

e (G

B )

Up to 15% CPU Reduction

Up to 24% Elapsed Time Reduction

½ the data to Recycle

92



z/OS 2.2 and SLB

� HSM apar OA 47876 for z/OS 2.2 fixes a problem in the HSM interface for SLB - BACKUP SYSTEM, RESTORE SYSTEM, and RECOVER. None of those utilities will work without the corresponding PTF applied.

93



z/OS DFSMS Highlights

� Catalog / IDCAMS– V2.2 GA

• Support for > 255 GDGs

• MODIFY CATALOG Security Enhancement

• Catalog RAS Enhancements

• IDCAMS VERIFY RECOVER

• IDCAMS RAS Enhancements 94



Catalog

� GDG Extended (GDGE)– Today, only up to 255 GDSs are controlled by a GDG.

– New enhancement: New DEFINE EXTENDED keyword allows up to 999 GDS to be active within 1 GDG.• DEFINE SMS.GDG GDG EXTENDED LIMIT(999)

– Default is NOEXTENDED LIMIT (255)– ALTER LIMIT(limit) is provided to modify the max number (between 1 and now 999)

• Enabled via SYS1.PARMLIB member IGGCATxx– New GDGEXTENDED variable set to “YES” (default is “NO”).

• Coexistence considerations– ALL systems in an installation should be at z/OS V2.2 or higher before enabling GDGEs, and it’s

expected that there will be no attempt to fall back a prior release. – Coexistence APAR OA46009 (Catalog) and OA46116 (DSS) prevents access to GDGEs on lower

level systems.

– SHARE Requirement Addressed: GO6SMG91037

�Why it Matters: Support a larger number of GDSs to make it easier to manage data sets that are created frequently. For example, it will be possible to specify a full year of GDSs be kept.

95



Catalog� Additional GDG Enhancements

– When users take the default NOSCRATCH option on DEFINE, it can leave unwanted data sets when GDSes roll off.

– In addition, users sometimes inadvertently put retention periods on their GDSs. System administrators would like for these data sets to be deleted regardless of retention periods.

– New enhancements:• New GDGSCRATCH (YES|NO) parmlib variable overrides IDCAMS keyword on

GDG DEFINE.– If not set or parmlib variable is GDGSCRATCH(NO), the keyword specified on the

DEFINE GDG takes precedence.

• New PURGE keyword added to IDCAMS DEFINE and ALTER GDG which allows user to specify what happens to unexpired GDSs during roll-off.– Specifies that when the SCRATCH is set, unexpired generation data sets (GDSs) are to

be deleted during roll off processing. Default is NOPURGE.– New variable, GDGPURGE (YES|NO), for SYS1.PARMLIB member IGGCATxx. Default

is “NO”. Overrides DEFINE GDG setting when PURGE is set.

�Why it Matters: Allows system administrators more control over and flexibility with GDG processing.

96



Catalog� MODIFY CATALOG Security Enhancement

– Existing requirements to provide access to the Catalog Address Space reporting commands without also allowing ability to issue commands which might alter the catalog configuration. • A user via SDSF might be able to issue an "F CATALOG,ALLOCATED" command,

but not be able to issue an "F CATALOG,CLOSE(usercat)" command.

– New enhancement: New RACF resource profile checked by catalog command processing to insure that the issuer of a command is authorized to execute that command.• Allows users utilizing SDSF to issue operator reporting commands, the results of

which they can view at their terminal, but not permit them to issue commands that change the catalog configuration.

• New optional OPERCMDS profile, MVS.MODIFY.STC.CATALOG.CATALOG.SECURE

�Why it Matters: Provide more granular security and better operational flexibility.

97



Catalog

� RAS Enhancements– DFSMSdss Restore of a User Catalog to Any

Volume • Today, DFSMSdss only allows a user catalog to be

restored to the same volume from which it was dumped (ie source dump volume must match restore target volume).

• New enhancement: Allows a user catalog to be restored to any volume.– Device geometry must match and applies only to

logical restore. – Restriction for physical data set restore remains.

�Why it Matters: Improves the users ability to recover a user catalog.

98



IDCAMS� RAS Enhancements

– REPRO MERGECAT output is too long• REPRO MERGECAT currently produces up to 6 lines for each entry that is processed;

one blank line, one or two for the IDC0639I message (Sphere conversion started), a blank line, and one or two for the IDC01402I (Sphere conversion ended). The one or two lines depend on the length of the dataset name.

• New enhancement: Add MESSAGELEVEL keyword to REPRO.

– New keyword MESSAGELEVEL(ALL| SHORT)– ALL is the default and is the same as the current

output– SHORT reduces the number of lines for each

entry

�Why it Matters: Satisfy customers requirements to make REPRO more usable and flexible.

• DB2 customers that move data catalog entry using MERGECAT can now see less messages and make the REPRO more manageable.

99



IDCAMS� RAS Enhancements

– PRINT and REPRO functions cannot process damaged VS AM CIs or records .• New enhancement: Add CIMODE keyword to PRINT and REPRO.

– New keyword CIMODE, which PRINTs or REPROs by CI rather than by record.

– Opens VSAM data set in CIMODE and is primarily for use with ESDS, but could be used to access a KSDS data component directly.

�Why it Matters: Improved error diagnosis / recovery by allowing processing if CI or VSAM record is damaged.

• DB2 customers have additional capabilities for error analysis.

100




� SMS

– V2.2 GA

• Space Constraint Relief

• RAS Enhancements

• Storage Group Space Alert Messages

101



SMS� Space Constraint Relief

– Currently for non-Guaranteed Space allocation, SMS Space Constraint Relief (SCR) processing reduces the requested space by the data class specified percentage, 'Reduce Space Up To (%)', if the originally requested space cannot be satisfied.

– New enhancement: SMS will extend the existing space reduction function to Guaranteed Space allocations to reduce allocation failures.• A new sub-parameter, Guaranteed Space Reduction (Y | N), is added to the data class

panel for the user to specify whether space reduction on guaranteed space allocation is permitted or not.

• Instead of allocating the smallest space quantity, SCR space reduction function is enhanced to allocate the largest possible space that satisfies the specified reduction percentage for both guaranteed space and non-guaranteed space requests.

• Note: Space reduction will remain unsupported for striping allocation.

�Why it Matters: Helps reduce allocation failures when the requested space is not available.

• DB2 customers should not use Guaranteed Space with the exception of the BSDS and active log data sets. If GS is used, it can now include a reduction of the space requested if necessary.

102



SMS

� RAS Enhancements– Read-only Variable

• Currently the user cannot provide installation specific values to the ACS routines via existing ACS read-only variables.

• New enhancement: SMS provides a new user defined ACS Read-only variable in SMS IGDSMSxx parmlib member:– USER_ACSVAR(value1,value2,value3)– The new parameter has 3 positional values.– Each value is a user-defined 8 character string. – The positional values of the new parameter can be altered by the following

SETSMS command: SETSMS USER_ACSVAR(value1,value2, value3)

�Why it Matters: Each installation can specify unique values, similar to how the system symbols are defined, to assign proper constructs without a need to modify the ACS routines constantly, making it easier to reuse ACS routines on different systems.

• Storage Administrators can now manage less environm ents, which can be especially helpful for DB2 customers.

e



SMS

� Read-only Variable Example

104

Production System

SYS1.PARMLIB(IGDSMSxx):USER_ACSVAR(PROD,,)

Test System

SYS1.PARMLIB(IGDSMSxx):USER_ACSVAR(TEST,,)

PROC DATACLAS /* DATA CLASS ACS ROUTINE */IF &USER_ACSVAR(1) = ‘ PROD' THEN

SET &DATACLAS = ‘PROD_DC’ ELSE

IF &USER_ACSVAR(1) = ‘ TEST’ THENSET &DATACLAS = ‘TEST_DC’

ELSE WRITE ‘***** UNEXPECTED VALUE FOR USER_ACSVAR’

EXIT CODE(0)END /* END DATACLAS ROUTINE */



SMS� Storage Group Space Alert Messages

– Currently, SMS VTOC Data Set Services (VDSS) issues IGD17380I to notify a user that the cumulative space allocated on the selected storage group has exceeded the high allocation threshold.• IGD17380I STORAGE GROUP (sgname) IS ESTIMATED AT xx % OF

CAPACITY, WHICH EXCEEDS ITS HIGH ALLOCATION THRESHO LD OF zz%

– New enhancement: SMS will be designed to allow you to specify new storage group space warning thresholds separately from the high allocation thresholds.• Create new alert threshold attributes in the storage group panel:

TOTAL SPACE ALERT THRESHOLD %...................... .. (0-99)TRACK-MANAGED SPACE ALERT THRESHOLD %.....(0-99)

• If the alert threshold value is not specified, it defaults to zero and no alert messages will be issued.

• New alert messages, IGD400I and IGD401I, will be issued to the console when the alert thresholds have been reached.

105



SMS� Storage Group Space Alert Messages

– IGD400I TOTAL SPACE ALERT ON STORAGE GROUP (sgname) CURRENT USAGE (xx%), ALERT THRESHOLD (yy%)

– IGD401I TRACK-MANAGED SPACE ALERT ON STORAGE GROUP (sgname)CURRENT USAGE (xx%), ALERT THRESHOLD (yy%)

• The alert messages will be issued at an incremental interval to prevent them from being issued too frequently.

• IGD401I is issued only for the pool storage group containing one or more EAV volumes.

�Why it Matters: Allows users to set a lower threshold for warning messages, which can provide more time to react to storage group space shortage conditions.

• DB2 customers can change the value for specific SGs in order to catch space related problems earlier. 10

6



SMS

� Storage Group Space Alert Messages– New enhancement: A new optional keyword, ALERT , is

introduced for the DISPLAY SMS command.• D SMS[,{STORGRP|SG}{(sgname| ALERT|ALL)}[,LISTVOL]• The new optional keyword, ALERT , is used to display any pool storage

groups which have already reached either the total space alert threshold or the track-managed alert threshold.

• The existing message, IGD002I, will be enhanced to include the space information: total space, total track-managed space, usage percentages and alert thresholds on the system at which the command is issued if the information is available.

• The space information is shown for pool storage groups only.

�Why it Matters: Makes it easier to see when it might be necessary to change a storage group's space management settings or add volumes to a storage group.



SMS

108




� Access Methods

– V2.2 GA

• RLS Index Level Locking

• Secondary Space Reduction

• LISTSTAT Command

• VSAM LSR Dynamic Buffer Addition

109



Access Methods

� RLS Index Level Locking– Today when processing VSAM data sets, there is a single point of contention

(DIWA) preventing concurrent VSAM RLS data CI splits in different CAs, CI reclaims, and spanned-record insertions and updates.

– New enhancement: Introduce a new level of serialization at the sequence set index CI level.• New locks used during CI SPLIT, CI RECLAIM, spanned record activity.

• CA split / reclaim activity will still use data-set level lock.• Enabled by default on all systems on V1R13 and above.

• New SMF fields were added to MVS System Management Facilities (SMF) for the new index record locks.

– Toleration Considerations:• Toleration APAR OA42676 is required to be installed on all V1R13 and V2R1 systems

across the sysplex prior to bringing a V2R2 system into the sysplex.

�Why it Matters: Reduces contention and improves per formance for many frequently updated VSAM key-sequenced and variable- length relative record (KSDS and VRRDS) data sets.

• Should help DB2 customers when processing many ICF Catalog records.

110



Access Methods

� RLS Index Level Locking– Additional feature will chain the I/O for the multiple segments of a

spanned record so that only one I/O is needed to write out the spanned record. • This will be available for NSR buffering only (not LSR or GSR).

�Why it Matters: Reduces I/O overhead when using spa nned records and the window for inconsistent spanned-rec ord segments.

111



Access Methods� Secondary Space Reduction

– Currently to allocate a secondary extent on the current volume, VSAM, RLS, PDSE, BAM and SAM will try to satisfy the user's specified amount on the DEFINE by calling DADSM extend services.• When there is not enough free space available on the current volume, EOV will call SMS to

switch volumes and try to extend onto a new volume.

• This algorithm can be inefficient in leaving large amounts of free space behind on the current volume which is less than what the user specified for secondary amount.

– New enhancement: Allow the system to extend data sets by less than their originally specified secondary space allocation amounts when doing so would avoid allocation

of space on additional volumes.– DB2 customers can see a reduction of volumes required for a

data set. • Keep in mind, once a data set was been allocated and extended to a new

volume, space on the original volumes are no longer eligible for that data set.

• For example, TS1 is allocated to VOL1 and then extended to VOL2 because of the lack of space on VOL1. You migrate most of VOL1 which now has a large amount of space. If VOL2 fills up, the data set cannot be extended back to VOL1 again.

112



Access Methods� Secondary Space Reduction

– DADSM will provide a new minimum allocation amount on the DADSM Extend interface.

• This new parameter along with the current secondary allocation amount on the IDCAM DEFINE, will allow DADSM the ability to determinate the largest free space to allocate.

– The allocated space must be equal to or greater than the minimum quantity derived from the attribute Reduce Space Up To (%) on the ISMF data class definition under “Space Constraint Relief”, and be less than or equal to the original specified secondary allocation amount.

• Reduce Space Up To means Remove up to that amount

– This support will be for SMS managed non-striped VSAM data sets and Non-VSAM data sets.

– New SMF fields:

• SMF64SSR –if secondary space reduction was used

• SMF64NTA –size of extent returned in Tracks

�Why it Matters: Help to improve disk space utiliza tion.

113



Access Methods

� LISTSTAT Command– Currently, to see the SMF64 data set statistics for a VSAM cluster, the data

set needs to be closed and then the LISTCAT command shows the latest SMF64 statistics.

– New enhancement: New function command to provide point-in-time SYSPLEX-wide statistics for open VSAM data sets that are using RLS.• IDCAMS SHCDS LISTSTAT(‘cluster’)• Does not require the data set to be closed to get the most current data set

statistics.

• Shipped on z/OS V1.13 and above with OA42435.

�Why it Matters: Improved reporting of VSAM RLS stat istics on demand and without the requirement to close the data set and then issue a LISTCAT command or retrieve SM F64 records. .

114




� Open/Close/EOV

– V2.2 GA

• Dynamic Tape Exits

• RAS Enhancements

115



OCE

� Dynamic Exits– Current design of DFSMS Open/Close/EOV Installation Tape exits requires IPL for

changes to be put into effect. • In addition, there is no way to add multiple exit routines for the exit.

– New enhancement: Several DFSMS Open/Close/EOV Installation Tape exits will be converted to be dynamic. • Impacted tape exits include: Label Anomaly, Volume Mount, File Validation, Fi le Start,

and File End .

• All exits are AMODE 31, reentrant and persistent until the next IPL.

• New rules for parameter list and execution, as well as return code decisions when multiple exits are called. See z/OS DFSMS Installation Exits, SC23-6850, for specific details and considerations.

�Why it Matters: Provides added flexibility, and no need to IPL to manage changes with Tape Installation exits. Allows multi ple exit routines to be associated with each exit.

116



OCE

� OCE RAS Enhancements– Abend code Elimination

• New enhancement: Eliminate abend code A13, reason code 18 by automatically selecting the correct volume during Open processing for multivolume tape data sets.

�Why it Matters: Provide improved usability.

– Add JOBID & SYSPLEXID to SMF 14/15• New enhancement: JOBID and SYSPLEXID are now added to SMF

14/15 (non-VSAM data set activity) records.– Rollback to V2.1 with OA43798.

• SHARE Requirement Addressed: SSMVSE11033 (SHARE Top 75)

�Why it Matters: Make chargeback and capacity planning easier.117




� DFSORT

– V2.2 GA

• zHPF Exploitation

• Functional Enhancements

118



DFSORT

� zHPF Exploitation– Currently DFSORT does not exploit zHPF.

• DFSORT normally uses EXCP for processing of basic and large format sequential input and output data sets (SORTIN, SORTOUT, OUTFIL).

• DFSORT already uses BSAM for extended format sequential input and output data sets (SORTIN, SORTOUT and OUTFIL). BSAM already supports zHPF.

– New enhancement: Update DFSORT to prefer BSAM for SORTIN/SORTOUT/OUTFIL when zHPF is available.• DFSORT will automatically take advantage of zHPF if it is available on

your system; no user actions are necessary.

�Why it Matters: Taking advantage of the higher start rates and bandwidth available with zHPF is expected to provide significant performance benefits on systems where zHPF is available.

119



DFSORT

� Functional Enhancements– Date conversion AGE function

• Customers often want to calculate the date duration that specifies the number of years, months, and days between an input date and current date.

• New enhancement: A date conversion function AGE for the BUILD and OVERLAY operands of DFSORT’s INREC, OUTREC and OUTFIL statements can now be used to calculate the date duration (ie time between a given date and the current date):– AGE=YMD produces a 8 byte result which has duration in years (0-9999), months (00-12), and days (00-31).– AGE=YM produces a 6 byte result which has duration in years (0-9999), months (00-12).– AGE=YD produces a 7 byte result which has duration in years (0-9999), days (00-366).

– Date conversion WEEKNUM function• Customers often want to calculate the week of the year for an input date.• New enhancement: A date conversion function WEEKNUM for the BUILD and

OVERLAY operands of DFSORT’s INREC, OUTREC and OUTFIL statements can now be used to calculate the week number that represents the week of the year.– WEEKNUM=USA returns an integer in the range of 1 to 54 that represents the week of the year. The week starts with Sunday,

and January 1 is always in the first week.– WEEKNUM=ISO function returns an integer in the range of 1 to 53 that represents the week of the year. The week starts with

Monday and includes 7 days.

�Why it Matters: Provide additional flexibility in creating reports and to help improve the usability of reports generated with these new functions.

120




� DFSMSrmm

– V2.2 GA

• WHILECATALOG Support

• SEARCHDATASET / SEARCHVOLUME Enhancements

121



DFSMSrmm

� WHILECATALOG Support– In V1R3 RMM introduced a new method for tape expiration processing.

• EXPDT (expiration date) allowed tape expiration to be on a specified date. This method is fast and simple, but does not allow expiration to be managed by catalog status (ie is the dataset cataloged).

– New enhancement: WHILECATALOG* attribute for datasets allows users to specify whether datasets should be retained or expired by date when catalog entries for them exist.

• Especially useful for GDG datasets – use WHILECATALOG to keep cataloged GDGs available until they’re uncataloged as new generations are created.

• Can be set as a default, or can be changed using CHANGEDATASET.

• Can be defined PARMLIB with separate defaults specified for GDG and non-GDG datasets.

�Why it Matters: Provides improved usability and fl exibility for tape retention / expiration processing.

122

*NOTE: Toleration PTFs are available



DFSMSrmm� SEARCHDATASET / SEARCHVOLUME Enhancements

– New enhancement: Search volumes and data sets by date/time ranges.

• SEARCHDATASE T and SEARCHVOLUME TSO commands are enhanced to allow users to select records by creation time.

• RMM SEARCHDATASET OWNER(*) LIMIT(*) CRDATE(START(2013/110,220000) END(2013/111,040000)) CLIST('RMM CHANGEDATASET ',')EXPDT(2013/120)')

– Date_Time_Range contains START(start_date,start_time )END(end_date,end_time )

�Why it Matters: Provides improved usability by nar rowing searches and creating smaller command lists.

123




� DFSMS / Storage Synergy– V2.2 GA

• Multi-Target PPRC

• IBM zHyperWritetm

• Easy Tier Application Hints

• XRC Workload Write Pacing

• Multiple Incremental FlashCopy

124



� Multi-Target PPRC (MT-PPRC)

– Allow a single volume to be the source for more than one PPRC relationship.

• The capability of having multiple relationships allows for many different possible configurations.

• This function will also support an incremental resynchronization capability between the two target volumes in both planned and unplanned HyperSwap® situations.

• TPC-R and GDPS will provide support for MT-PPRC.

�Why it Matters: Having a second PPRC target and allowing for a quick incremental resync between the two targets prevents losing all Disaster Recover capability from a single outage.

Storage Synergy

125

h1

h2

h3

Metro

Mirror

Metro Mirror



Storage Synergy� IBM zHyperWrite tm - DS8870 Log Write Accelerator for DB2

– New technology that combines DS8000 and z/OS enhancements that deliver performance benefits for writing operations to DB2 logs in the Metro Mirror (PPRC) environment. • Reduces latency overhead compared to normal storage based synchronous mirroring.• Can help reduce up to 43% of the DB2 log write time.*

�Why it Matters: Provide improved DB2 transactional response time and log throughput improvements.

� Easy Tier Application Hints– System Storage Easy Tier is now designed to allow software-defined policy

information to be communicated to Easy Tier control units, to help them efficiently deploy storage. • Easy Tier currently optimizes data placement and tiering based on workload history and this

does not always reflect the future performance requirements of the data.• New planned exploitation by DB2 to provide hints to Easy Tier on the DS8000 when a

database reorganize is done so that the target data set for the database will be reorganized and optimized based on the activity of the application.

�Why it Matters: Intended to help steer data placement within Easy Tier volumes to meet application performance objectives.

126

*Disclaimer: Based on projections and/or measurements completed in a controlled environment. Results may vary by customer based on individual workload, configuration and

software levels. 1 Measurements for comparisons were completed as part of a formal performance evaluation on a dedicated, isolated test system.



Storage Synergy

� XRC Workload Write Pacing – XRC is designed to work with z/OS Workload Manager to control XRC Write Pacing.

• Reduces administrative overhead on hand managing XRC write pacing.

• Avoids the need to define XRC write pacing on a per volume level allowing greater flexibility in configurations.

• Prevents low priority work from interfering with the Recovery Point Objective of critical applications.

�Why it Matters: Simplifies XRC usage by eliminating the need to constantly adjust the write pacing settings and/or allocation of data to volumes..

� Multiple Incremental FlashCopy– New FlashCopy function that supports up to 12 targets for incremental FlashCopy.

• Previously only a single incremental FlashCopy was allowed for any individual volume.

• Incremental FlashCopy allows a full physical copy to be created much more quickly than a non-incremental FlashCopy.

�Why it Matters: Provide more flexibility and resilience for FlashCopy helps to better protect application availability and provide improved data protection across physical volume failure events.

127

db2 ats (advanced technical support) -...

Documents