Mainframe batch monitoring
Batch monitoring that says when a batch will be late.
Meridian monitors every batch job on every z/OS LPAR: what is running, waiting, failed and late, with the scheduler's own view beside it. Hours before the deadline, it says in plain words whether each batch will meet its SLA, and puts the fix one click away.
For service providers running many clients and for enterprises running their own estates. A free 90-day proof of concept on one LPAR.
Example data from a lab estate.
What is monitored
Everything a batch depends on, on one screen.
A batch is late because of a job, a WTOR, a subsystem or a full initiator class. Meridian watches all of them together, per LPAR and per client, so the cause is on the same screen as the delay.
Jobs and queues
The scheduler's queues and today's runs: each job running, waiting, done or failed, when it started, and when it usually ends on this weekday. CA 7 today.
Every abend, with its step
Each job abend with its code, the step, the scheduler's view of the job and what was done about it, linked to the console line that reported it.
WTORs holding work
Every outstanding WTOR at the top of the console with its age, so a job waiting on an answer is seen in minutes, not at the deadline.
The subsystems a batch needs
CICS regions, IMS, Db2 and Adabas, up or down, live. A batch that needs a region that is down is predicted as such.
Initiators and spool
Initiators active and idle, and spool used, with history per LPAR, so a full initiator class or a filling spool is seen before it holds the batch.
Named batches, in words
Every batch you name, with its deadline and its state: On time, On time, but tight, Slightly delayed, within the approved time or SLA breach incoming.
From monitoring to prediction
Not just late. Late, hours before.
Monitoring tells you a job failed. Prediction tells you whether the batch can still make it. Meridian does both, from facts, not guesses.
- The chain, from your scheduler's own definitions. Nobody draws it by hand.
- Your own run history: the usual run time of each job by weekday, with its spread.
- The live state of the estate: the subsystems and initiators each job needs.
- Recalculated the moment anything changes. Until there are a few weeks of history, Meridian says it is not sure, in those words.
From an alert to the fix
The fix is one click away.
An alert that only says something is wrong leaves the operator to open three more tools. In Meridian the alert and the action are the same card.
- What Meridian saw, and why it proposes the action.
- The scheduler command, exactly as it will be sent.
- Whether the batch can still end on time after a restart, and by how much.
- The outcome on the same card: what came back, when, and who confirmed it.
Every scheduler action
Restart from a step, force complete, cancel, hold, release, demand, run now, requeue, and the queue-wide holds and releases, each from a card with one confirm click.
Alerts you set
Rules on message ids, job names, abend codes, thresholds and SLAs, sent to the channels you choose. Meridian's own alarms go to a separate channel.
Reports
Attainment by month (how many batches ended on time) and prediction accuracy (what Meridian said an hour before, and what happened), per client.
Who it is for
Built for the floors that answer for the batch.
- One install, every client, every LPAR
- Each client's batches, rules and people its own
- Attainment and prediction accuracy per client
- Batch deadlines seen hours early
- Every action recorded for your auditors
- Runs inside your network; your data never leaves it
Questions
Questions about batch monitoring.
Short answers. Or write to hello@hexab.in.
How to evaluate a batch monitoring toolWhat is mainframe batch monitoring?
Watching the batch jobs that run on z/OS, usually overnight, to see which are running, waiting, failed or late, and acting before a batch misses its deadline. Meridian adds the step most monitoring stops before: it predicts, in plain words, whether each named batch will meet its SLA.
How is this different from the scheduler's own monitoring?
Scheduler tools watch their own jobs and predict from run history with labels such as on time or late. Meridian also takes live subsystem status into the prediction, shows the console, WTORs and tickets on the same screen, answers on the restart card whether the batch can still make it, and works across many clients.
Which schedulers does Meridian monitor?
CA 7 today, read and driven. Other schedulers are on the roadmap. Tell us what you run. The console, dashboard, health and capacity views work without a scheduler connection.
Does batch monitoring need agents on the mainframe?
Meridian has a light footprint on the mainframe and is installed with your team, inside your network; your data never leaves it. Your security team decides what it may see and do, and our engineers go through the details with your systems programmers before anything is installed.
Can it monitor batches for many clients at once?
Yes. One install serves a whole company. Each client's batches, deadlines, rules and people are its own, and every screen says which client is shown.
What does it cost?
Per LPAR per month, billed yearly. Watch, at $150, includes the batch views, health, capacity and alerts. Act, at $500, adds SLA prediction and reports, actions from cards, Builder flows and the ticket tool. There is a yearly company fee of $40,000, no per-user charge, and a free ninety-day proof of concept on one LPAR. Pricing.
See it on your own estate.
A ninety-day proof of concept on one LPAR, free, with our engineers on the call. After ninety days you decide on your own results.