Computer Corporation of America
|
Feedback
Search CCA:
   
USA CCA
CCA Products
CCA Customer Support
CCA Resources
CCA - Company
CCAPRINT: A Newsletter for Model 204® and System 1032® Users
September 10, 2005
     
Improving Recovery Performance with CCAJLOG Printer-friendly version
Model 204 V6R1.0 Documentation Refresh  

Model 204
USE OF AND ACCESS TO PRODUCTS AND FEATURES ARE IN ACCORDANCE WITH THE TERMS AND CONDITIONS OF THE USER’S SOFTWARE LICENSE. THE PRESENTATION OF MATERIAL HEREIN DOES NOT, IN ANY MANNER, MODIFY SUCH TERMS AND CONDITIONS.

Improving Recovery Performance with CCAJLOG
By James Damon



During a production Online run, roll forward recovery information is written to the CCAJRNL dataset. Additionally, a substantial amount of unformatted CCAAUDIT data is also written. When recovery is required and after roll back recovery has completed, that dataset (during recovery its ddname is CCARF) is read and the roll forward information is extracted and used to reapply all committed updates up to the time of the system failure. However, the CCAAUDIT information must also be read, although it is ignored since it has no value to roll forward recovery. This can slow down roll forward recovery, if there is a substantial percentage of CCAAUDIT information or, if many minutes elapsed between the last checkpoint taken and the system failure.

In V6R1.0 and all operating systems, this problem has been mitigated by the introduction of CCAJLOG, an optional new ddname and dataset. If you allocate this ddname to the run, the unformatted CCAAUDIT data is written to CCAJLOG and only roll forward recovery data is written to CCAJRNL. This improves recovery performance and also reduces the likelihood of filling CCAJRNL, which results in run termination.

CCAJRNL before V6R1.0
In the Figure 1 schematic, all data written to CCAJRNL, including unformatted CCAAUDIT data, is written from CCAJRNL buffers allocated in the address space. The number of buffers allocated is determined by the NJBUFF parameter.

Figure 1. CCAJRNL set up to collect all roll forward and audit information

Introducing CCAJLOG
In V6R1.0, if you allocate the CCAJLOG ddname to the run, all unformatted CCAAUDIT data is written to CCAJLOG.

You can now produce a formatted CCAAUDIT report from CCAJLOG using the AUDIT204 utility. Prior to V6R1.0, this was accomplished by running AUDIT204 against CCAJRNL.

Figure 2. CCAJRNL collecting roll forward data; CCAJLOG collection audit data

CCAJLOG and CCAJRNL Datasets as Input to Utilities
When you need a CCAAUDIT report, use the CCAJLOG dataset as input to the AUDIT204 utility, under the ddname of CCAJRNL. You can also use CCAJLOG as input to the UTILJ utility, again under the ddname of CCAJRNL; however, the only useful report produced is a histogram of the number of blocks and block sizes that were written to the CCAJLOG dataset. CCAJRNL continues to be the input dataset provided to RESTART, for roll forward recovery (ddname=CCARF); to REGENERATE, for media recovery (ddname=CCAGEN); and to UTILJ, (ddname=CCAJRNL).

Figure 3. Utilities using CCAJLOG and CCAJRNL datasets as input

Managing Space Requirements for All Operating Systems
The space requirements for CCAJRNL are reduced, if CCAJLOG is enabled. However, CCAJRNL must still be sized to prevent B37, D37 or E37 abends, which will terminate the run and require recovery to recover files to a state of logical and physical consistency. The same applies to CCAJLOG: if it fills, the run terminates and recovery is required. Generous secondary extents allocated to these datasets also help avoid overflow problems.

Additional Managing Space Requirements on z/OS
At z/OS sites, the most comprehensive way to avoid dataset full problems is to define both CCAJRNL and CCAJLOG as generation data groups (GDGs), which is also a new feature in V6R1.0. At this time see the Model 204 System Managers Guide V6R1.0 for further details regarding GDGs. I’ll discuss GDG’s in greater detail in subsequent articles.

Summary
If you are looking for ways to maximize recovery performance and to reduce the risk of filling CCAJRNL, the new features - CCAJLOG and GDG’s - available in V6R1, will help achieve those objectives. No changes to CCAIN parameters are required to implement CCAJRNL splitting. The presence of CCAJLOG in the JCL or as a DEFINE STREAM in CCAIN will enable the feature.

Model 204 V6R1.0 Documentation Refresh
CCA is pleased to announce that the Model 204 Documentation Library CD was refreshed at the end of August 2005. If you picked up your copy of the Model 204 Documentation Library CD at Insight 204 in June or you ordered V6R1.0 before the end of August, please contact CCA so we can send you the more recent library CD. If you are accustomed to downloading individual documents from the CCA web site, rest assured that they are the refreshed editions. Any orders for V6R1.0 placed the last week of August or later will receive the more recent library.

 


Contact CCA Webmaster
Copyright 2008