JAMS Scheduler for Complex Batch Processing

Batch processing rarely fails at the job that breaks. It fails three jobs later, when a report goes out built on yesterday’s data because nothing told the downstream step to wait.

Picture a typical nightly close. A bank file lands on an SFTP server. An ERP posting job runs on Windows. A Python transform runs on Linux. A SQL Server load and a BI refresh follow. Often each step has its own scheduler: Task Scheduler on one server, cron on another, SQL Server Agent on the database, a native scheduler inside the ERP. Each one knows when its own job ran. Do any of them know whether the previous job finished?

Workload automation closes that gap by putting every step under one scheduler, with a shared dependency model, alerting, and run history. JAMS Scheduler runs that kind of chain from a single console, on the platforms where the work already lives.

Pricing is published on the website, so you can see it without a sales call. JAMS Scheduler Core starts at $833 per month, billed annually. JAMS Scheduler also offers per-agent pricing, and agent-based licensing includes unlimited executions, so adding jobs to the schedule doesn’t require add-ons or a plan upgrade.

One workload automation layer for hybrid batch processing

JAMS Scheduler installs on Windows and uses Microsoft SQL Server as its database. Job definitions, calendars, and run history all live in that one place. Agents go on the machines that do the work: Windows, Linux, UNIX, IBM i, z/OS, and OpenVMS.

That matters for complex job scheduling because a batch chain is only as reliable as its weakest handoff. When the Linux transform and the Windows ERP job report to the same scheduler, the dependency between them lives on the job: JAMS Scheduler holds the downstream job in a waiting state until the upstream job finishes with the completion status you require.

Agents can be grouped into pools and queues, so work routes by priority and capacity instead of to one named server. Integrations cover PowerShell, SQL, Azure Data Factory, and Python. JAMS Scheduler also has native SAP and JD Edwards job types.

JAMS Scheduler imports existing jobs from cron, SQL Server Agent, and Windows Task Scheduler. If you’re moving off another workload automation tool, JUMP, the migration tool JAMS built for JAMS Scheduler, converts competitor job definitions across nine platforms and automates roughly 75% of the conversion. JAMS Scheduler also runs on AWS or Azure.

Event-driven scheduling for complex job dependencies

A fixed start time is a guess about when the data will be ready. Event-driven scheduling replaces the guess with the event itself.

JAMS Scheduler has six trigger types: schedule, interval, job completion, file, variable, and email. A job completion trigger can fire on success, on failure, or on any completion, so a recovery job can start the moment its parent fails instead of waiting for the next window. Business calendars accept natural language rules.

File triggers deserve a closer look. A file trigger can watch for a file to be present, absent, or available, with an optional minimum file size as a check against empty or truncated drops.

Triggers start work. Dependencies gate it. A trigger decides when a workflow begins; a dependency is a prerequisite checked once the job is on the schedule. If a dependency isn’t met, the job waits until the job it depends on finishes with the completion status you require.

Inside a workflow, conditional branching and error handling decide what happens after each step, and visual job maps and relational views show the chain in a single console. Parameters and variables let one job definition serve several environments or data sets, so the nightly close for each region doesn’t need its own copy.

Monitoring batch job orchestration before the SLA slips

The useful alert is the one that arrives while there’s still time to act.

JAMS Scheduler lets you set SLA windows with thresholds that forecast risk, so a job running long gets flagged before the deadline passes, not after. The monitor shows queues, agents, and jobs live from one console, alongside visual job maps and relational views for triage.

When something does fail, retries with backoff, skip, and rerun handle the common cases automatically. Recovery jobs take over based on the job’s completion status, and notifications route to the job’s owner. For environments where the scheduler itself can’t go down, high availability and fault-tolerant configurations are available.

JAMS Scheduler Core includes 24x7 support. The support team uses round robin rather than tiers, senior engineers take new tickets, the team sits in the US, the UK, and Australia, and support is never outsourced.

Audit-ready control for enterprise workload management

An auditor’s first question about a batch process is usually who changed it. With jobs spread across native schedulers, the honest answer is a search through several servers’ logs.

JAMS Scheduler keeps an audit trail of the changes made in the system. Permissions on users and Active Directory groups control who can see and edit each job, and SSO, SAML, MFA, approvals, and separation of duties are supported.

Two controls do most of the work in practice:

• Credentials stay in the platform. Jobs reference a stored credential instead of carrying a password in a script, and data is encrypted in transit and at rest.

• Changes move through promotion. Job definitions promote from development to QA to production with approvals along the way.

Complex batch processing in production

Stater, an independent mortgage services provider, ran its core batch on OpenVMS, its Windows jobs in Task Scheduler, and its database work in SQL Server Agent. Different groups ran their own jobs, and there was no central record of when a job ran, whether it succeeded, or who owned it when it failed. After the move, Sander Vos of Stater put it this way: “JAMS is centrally managing every batch process that is run across our organization.”

At Matlock Capital, a Chicago trading firm, the problem was a set of interdependent overnight jobs that post-process market data for the next trading day, running on several operating systems in more than one location. Operations engineer JP Layug said the setup was straightforward and “the results are very reliable.”

Some teams don’t get a long runway. Kent State University had 26 days between contract award and the expiration of its old scheduler, and rebuilt all 182 of its Banner job chains in that window.

See JAMS Scheduler on your own batch chain

The fastest way to judge a scheduler is to put your ugliest dependency chain in it. Download the JAMS Scheduler free trial and build that chain in your own environment.

Related reading: Workload automation with JAMS Scheduler · Enterprise job scheduling