AutoSys, Stonebranch and other schedulers
Not every site uses Control-M or IBM. AutoSys and Stonebranch are common in banks, and older mainframe shops may run CA 7, ESP or Zeke. The ideas are the same; the words and commands change.
AutoSys Workload Automation
AutoSys, now owned by Broadcom, defines jobs in a text language called JIL (Job Information Language). Jobs can be grouped in box jobs, and dependencies are written as conditions such as success(LOADTXN).
insert_job: UPDACCT job_type: CMD box_name: PAYROLL_NIGHT command: /apps/pay/bin/updacct.sh machine: payhost01 owner: payops condition: success(LOADTXN)
| Command | What it does |
|---|---|
| autorep -J UPDACCT | Report a job's status and last run |
| sendevent -E FORCE_STARTJOB -J UPDACCT | Start a job now, ignoring its conditions |
| sendevent -E JOB_ON_HOLD -J UPDACCT | Put a job on hold |
| sendevent -E JOB_OFF_HOLD -J UPDACCT | Release it |
| sendevent -E JOB_ON_ICE -J UPDACCT | Put on ice: skip it, and treat it as satisfied for dependants |
Stonebranch Universal Automation Center
Stonebranch UAC combines a web-based Universal Controller with Universal Agents on each system, including z/OS. Work is defined as tasks, chained into workflows, and started by triggers such as a time, a file arriving or an API call. It is often chosen to modernise or consolidate older schedulers.
Other names you may meet
| Product | Where you see it |
|---|---|
| CA 7 Workload Automation (Broadcom) | Long-established mainframe scheduler in many banks |
| ESP Workload Automation (Broadcom) | Mainframe and cross-platform scheduling |
| Zeke (Rocket Software) | Mainframe scheduling in some sites |
| Redwood RunMyJobs | SaaS scheduling, often with SAP |
| Apache Airflow | Open-source workflow scheduling for data pipelines |
Translating between tools
| Idea | Control-M | TWSz / IWS | AutoSys | Stonebranch |
|---|---|---|---|---|
| Group of jobs | Folder | Application / job stream | Box | Workflow |
| Dependency | Condition | Predecessor / follows | condition: | Workflow link / task monitor |
| Daily plan | Active jobs (by ODAT) | Current plan / Symphony | Event server | Activity |
| Command line | ctm commands / Automation API | conman, batch utilities | autorep, sendevent | Command line interface and REST API |
Common mistakes
Force-starting ignores the reason the job was waiting, such as a missing file. Find out what it is waiting for first.
Its dependants may run immediately with missing data. Use hold when you want everything downstream to wait.
A 'box' in AutoSys, a 'folder' in Control-M and a 'job stream' in IWS are similar, not identical. Check the tool's own rules for restarts and dependencies.
What you will see at work
- Job adverts often list several schedulers. Experience in one is usually accepted as a strong base for another.
- During migrations, sites run two schedulers side by side and link them, so you may need to trace a dependency across tools.
- Scheduler definitions are increasingly kept as code (JIL files, JSON) in Git.
Key terms
Check your understanding.
Take this lesson's quiz and save your progress. Free.