JES3plus: The Time is Nigh

Phoenix Software explains why IBM JES3 clients should consider migrating to JES3plus

If you are an IBM JES3 shop considering an upgrade to z/OS 3.1, then you have a decision to make. With z/OS V2R5 being the last version of z/OS to include JES3, should you embark on a wholesale conversion to IBM JES2, fraught with potential financial, business and operational risks, or perform a trivial migration to Phoenix Software International’s JES3^plus and breathe a sigh of relief?  How do you navigate the critical data points that you must consider before making this decision? We’re here to help you review the options and make the right decision for your organization.

The Technology

For those less familiar, the Job Entry Subsystem (JES) is a component of IBM’s z/OS operating system responsible for managing batch workloads. Minimum JES functionality includes reading jobs into the system, selecting them for processing, managing/printing their output and purging them from the system.

Uniquely, JES3 provided:

• Multisystem coupling
• Intelligent tape sharing
• Resource routing
• Load balancing
• Dynamic initiators
• Pre-execution device setup
• Job networks
• Deadline scheduling
• Multisystem consoles
• A JESplex-wide merged log
• Other capabilities that simplified the operation and management of loosely-coupled systems nearly two decades before the emergence of sysplex support in the base operating system

These and other unique features, along with superior SPOOL I/O technology and performance, made JES3 the favored alternative in many client installations, especially those with high-volume data processing requirements. JES3 was considered the premium JES choice and IBM assessed a nominal license upcharge for its use.

What Is JES3^plus?

In February of 2019 IBM released a Statement of Direction announcing that the release following z/OS V2R4 would be the last to offer JES3 and that began a countdown to a final JES3 sunset. At the urging of the loyal JES3 client base, Phoenix and IBM negotiated a source code license for the JES3 technology with the intent of creating a Phoenix-owned derivative product. JES3^plus is that licensed, derivative work. JES3^plus includes all the base IBM JES3 functionality plus important enhancements.

JES2 Conversion Versus a Trivial Migration to JES3^plus

Let’s explore the cost benefit associated with these two choices. Some clients already made the decision to move to JES2, some performed an analysis in an attempt to determine the hard and soft costs of conversion, and others have yet to do anything. Which of these describes you?

Shops with hands-on conversion experience tell us that it is not as straightforward as some might think. There are thousands of little details. One large insurance carrier that opted for a JES2 conversion prior to the JES3^plus announcement said, “We wish we would have known that there was going to be an alternative. We would not have done the conversion.” Another financial institution, who studied the conversion costs and chose to remain with the JES3 technology, told us their ROI would have taken 25 years.

Obviously, the size and complexity of your organization and its workloads play a major role in the time, effort and associated costs of conversion. In almost all cases, it is a long, intensive process that can cost hundreds of thousands, if not millions, of dollars. Most organizations do not possess the resources to take on this type of project and even those that do have a hard time justifying a project to prioritize this core technology infrastructure change over other business needs. Particularly when, depending on how many JES3-unique features they leverage, they might even be taking a step back. While JES2 has been enhanced to add in JES3-like functionality, the reality is that the core technology design remains unchanged. Let’s be honest: If you have not already done the conversion to JES2, or have at least gotten most of the way through it already, you have waited too long.

Let’s contrast that with a trivial migration to JES3^plus. Why do we use the word “trivial?” Because the migration from IBM JES3 to JES3^plus is typically no more effort than implementing and testing the IBM JES3 included in a new release of z/OS and is completely transparent to your end users. That means no additional projects added to the docket or the budget. And, ongoing license costs for JES3^plus are typically the same or less than IBM’s z/OS optional license charge for JES3. Business as usual, no upheavals.

One large JES3^plus client purposely said nothing to their users about their JES3^plus migration because they wanted to see if their users would observe any differences. They were happy to report that no one noticed anything had changed. Their POC and implementation were uneventful and “as advertised.”

What Does a Trivial Migration to JES3^plus Look Like?

Well, first it is important to note that this migration should preferably be done while you are still on a version of z/OS that has IBM JES3 (standard disclaimer). While rollback is unlikely, your compliance folks will appreciate the fact that you have a rollback option available.

It is also important to understand that we have taken great care to ensure that JES3^plus and IBM JES3 can co-exist in the same environment. Some clients have leveraged this to do a phased rollout while others have just taken a “Big Bang” approach.

Here are the high-level steps:

1. Install JES3^plus and cumulative service in new or existing SMP/E CSI and zones.
2. Install user exits and other modifications just as you would for IBM JES3.
3. Recompile anything else that uses JES3 control blocks—again, just as you would for IBM JES3.
4. Remove //STEPLIB DDs from JES3 and JES3CI JCL procedures.
5. Bring up JES3^plus exactly as you would a new IBM JES3 release via IPL.
6. Test your applications—they won’t notice the difference.

When a phased rollout approach is used:

• Rolling IPLs allow JES3^plus to replace IBM JES3 one LPAR at a time in an existing JESplex (there are no restrictions on LPAR re-IPL order)
• A JES3^plus global can DSI to an IBM JES3 local for global takeover and vice versa
• IBM JES3 does not understand the new features added to JES3^plus, so don’t use them until your phased rollout is complete

How Phoenix Has Improved the JES3 Technology

One of our first orders of business was to form the JES3^plus Client Advisory Council (JCAC) to assist us in prioritizing enhancements that would provide real business value to our clients. Through annual releases (versus IBM’s historically biennial releases) and service, we have seen a steady stream of significant enhancements, including:

Functional Enhancements

• WLM batch initiator balancing
• Dynamic system affinity changes
• Dump Job (DJ) filtering
• JCL job card passphrases
• SMF 84 and IEFUSO enhancements
• Support for Job Correlator—used primarily by z/OSMF
• Support for the SEGMENT= and SPIN= JCL keywords on individual SYSOUT DDs
• More efficient Job Submit processing
• More efficient SPOOL utilization by ASR control blocks
• z/OSMF installation workflow (along with continued support for the traditional path)
• And more

Performance Enhancements

• SPOOL I/O now supports ACKD, zHPF and PAV
• Reduced Local Lock Contention

Enhancements In Development

• JES Policy
• WLM AI batch initiator support
• Modify Job SSI (SSI 85)—used primarily by z/OSMF
• Dynamic Duplicate Logon Change to help implement z/OSMF

SPOOL I/O Performance Benchmark Results

Let’s explore the important SPOOL I/O performance enhancements that we have made a bit further.

Observations

• Increased overall throughput and improved MVS dispatching efficiency
• Nearly twice as many jobs completing while reducing channel utilization by a whopping 92%

Enhancement Details

• Support for newer channel programming techniques
  • Advanced CKD to leverage modern RAID DASD I/O optimizations when zHPF is not available
  • High Performance FICON (zHPF) for significantly higher I/O start and data throughput rates
• Support for Parallel Access Volumes (PAV)
  • Multiple I/O launched against a single SPOOL extent instead of always chaining multiple requests
  • Clients control the level of parallelism via a new MAXPAV= specification on the OPTIONS statement
• Reduced Local Lock Contention
  • Allows more SPOOL I/O operations to complete simultaneously
  • Reduces the total number of I/O-related SRBs scheduled

Benchmark System Resources and Methodology

For detailed information about this benchmark, including available hardware resources and testing methodology, refer to this document.

Which Direction Will You Go?

If you are an IBM JES3 shop considering an upgrade to z/OS 3.1, then the time is nigh and a choice must be made.

Why incur the risks and costs of a wholesale JES2 conversion when you have a truly viable alternative? Instead, perform a trivial migration to JES3^plus and enjoy full investment protection for all of the money, time, and effort you’ve spent building a business around JES3 features and functions, thus fulfilling the value proposition of the mainframe. Use your time, human resources, and hard-dollar savings to focus on your business objectives and not your infrastructure.

Want More Information?

If you are interested in more information on Phoenix Software’s Spooler offering, check out our website and request additional information. We would love to chat with you about next steps and listen to any input you may have on the future of this product.

---