On Feb. 9, IBM announced the release of mainframe operating system z/OS 1.12, which the company plans to make available...
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
by September of this year.
For some inexplicable reason, the first bullet item at the top of IBM's z/OS 1.12 announcement trumpets the ability of Predictive Failure Analysis to detect when a system is writing too many System Management Facility (SMF) records. I don't know about other shops, but this hasn't been a problem for us in a while. Fortunately, deeper down in the announcement, we find some substantive changes that make z/OS 1.12 a worthwhile release.
Z/OS 1.12 and Data Facility
IBM takes an unglamorous yet very important step forward by broadening support for extended address volume (EAV) disks. EAVs contain extended addressing space, which goes beyond the old architected 65,520-cylinder limit. This is IBM's strategy to keep the number of devices fixed at 6,400 while making individual volumes larger.
Jumping into the EAV parade with z/OS 1.12 is just about every access method you can think of, including old ones like basic direct access method (BDAM) and the newer PDS/E. A bunch of subsystems and utilities gain this support as well, including JES2 and JES3, which will be able to use the behemoth volumes for checkpoint and spool volumes.
It looks like EAV support is nearly universal now, so shops have few excuses for not moving onto larger volumes. Also note that, in addition to EAV support, the updated access methods will be able to use the extended task I/O table and data set association blocks above the 16 M line.
Two items will impact VSAM. First, IMBED and REPLICATE, two attributes IBM has been trying to kill for years, will not be preserved for clusters dumped with DFDSS or HSM. Second, VSAM will dynamically reclaim Key Sequenced Dataset (KSDS) control areas (CAs) left empty from previous split activity. The hope is that defragging the cluster on the fly will reduce offline re-orgs. Time will tell us how well this works.
Finally, z/OS 1.12 has a new way of naming temporary data sets in case you have two jobs with the same name starting at the exact same second -- it could happen.
Security in z/OS 1.12
The new MVS will include the Elliptic Curve Cryptography (ECC) algorithm, which is blessed by the NSA as being faster, with smaller keys. IBM quickly included ECC as a choice for nearly anything that needs to be encrypted or decrypted.
In addition, z/OS 1.12 includes many significant enhancements for public key infrastructure. Foremost are some utilities aimed at certificate management and support for Certificate Management Protocol. There are a number of added usability features, including Resource Access Control Facility's ability to send an e-mail detailing why it rejected a certificate.
Z/OS 1.12 introduces the interesting idea of "trusted" TCP connections. From the announcement, this sounds like an extension of the old LU6.2 security model, where two endpoints that trusted each other needed only to exchange user IDs for authorization. Trusted connections work similarly, except both ends have to be in the same Sysplex and security domain. This type of authentication should be much cheaper than Secure Sockets layer, although the messages will not be encrypted. I also doubt trusted connections will be allowed off-platform.
IBM continues to enhance big iron's availability. A new item will be Run Time Diagnostics, which is supposed to detect bad things and gather the necessary information to reduce the time humans need to identify the problem. These facilities are in the very early stages of development but it will be interesting to see how they evolve.
Cross-system Coupling Facility (XCF) will watch for dead or dying LPARs and, if necessary, take actions to vary them out of the Sysplex. There are also measures to detect when a coupling facility (CF) structure takes too long to respond.
The capacity-provisioning component will use IMS and CICS monitoring data when considering overall performance. It will take actions necessary to bring new resources to bear if it thinks either transaction processor may start missing its response time goals.
Other z/OS 1.12 tidbits
Other interesting items:
- The program management binder will work better with the existing binder C and C++ DLL functions to avoid much of the overhead incurred when switching between XPLINK and non-XPLINK programs. This might help some shops with mixed-language applications.
- Enhancements to the display XCF ("D XCF") command show the possible consequences of a CF allocate command. There's also a REPORT parameter designed to show the results from the previous REALLOCATE.
- Z/OS 1.12 SDSF's programming interface will include some Java classes.
- A utility to test the integrity of PDS/E datasets
- New macros IXCMSGOX and IXCMSGIX support 64-bit XCF message buffers. No more copying messages back and forth across the bar.
- The ability to store file format (e.g., LRECL and BLKSIZE) in Unix System Services files. This makes them much easier to access from regular z/OS programs.
- The Web-based z/OS Management Facility (z/OSMF) expands to include Workload Manager policy editor. Shops will also be able to add non-z/OS MF application launch points.
There are too many other single enhancements to mention here. My advice is to look in the announcement for your pet peeve to see if z/OS 1.12 addresses it.
Your take on the mainframe
What is your favorite thing about the mainframe? Let us know what you think makes big iron great. It can be anything from "rock solid security" to "IBM had to create a fix for IEFBR14."
ABOUT THE AUTHOR: For 24 years, Robert Crawford has worked off and on as a CICS systems programmer. He is experienced in debugging and tuning applications and has written in COBOL, Assembler and C++ using VSAM, DLI and DB2.
What did you think of this feature? Write to SearchDataCenter.com's Matt Stansberry about your data center concerns at email@example.com.