Product · Operations
Sentinel signs in, raises a work order and syncs a handheld, every few minutes, all night.
Two jobs sit over a Maximo estate: watch the platform, and exercise the work. Sentinel takes the second. When one of the five journeys stops completing, the alert names the journey, the failing step, the environment and when it last worked.
Two jobs over one estate
Watching the platform and watching the work are different jobs
Infrastructure monitoring
Keeps the platform inside its limits. Necessary, and worth keeping.
Maximo journey monitoring
Confirms the work still completes, on each environment, minute by minute.
What it exercises
Hosts, containers, storage, network, certificates, database sessions, queue depths.
Sign-in, work order creation, an integration message, the scheduling view, a handheld sync.
What a green board proves
Nothing in the platform is outside its thresholds right now.
The journey ran to completion inside its expected time, on that environment, minutes ago.
What it gives you at 04:12
A resource graph, and a correlation exercise across tools before anyone can state scope.
The journey that failed, the step, the environment, and the last time it passed.
What it hands over
A metric that has moved, and an engineer deciding whether it matters.
An evidence bundle Diagnose picks up to establish the cause.
The left column, running alone
What the platform job leaves uncovered
The first alert is a phone call
A supervisor cannot save a work order, rings a colleague, who rings the service desk. By the time anyone opens a screen the failure is twenty minutes old and third hand.
Seven tools, none holding the question
Hosts, containers, database, network and storage each read green. None of them answers whether a supervisor can raise a work order right now.
Monthly availability is computed from host uptime
Monthly service reporting uses host uptime, which is true about servers and a poor proxy for whether people on shift could work.
The overnight window is thinnest
Where journey checks exist they are run by hand in office hours. Batch windows, patches and partner changes land when nobody is looking.
The first hour of an incident goes on proving something is wrong, and the service report says available while the depot remembers working on paper.
The five journeys
What a failure of each one costs the operation
You name the journeys you cannot lose, and the cadence each is checked on. Watching Maximo and MAS estates MaxIron operates 24/7 across utilities, transport, oil and gas, manufacturing and maritime, default in MaxIron managed hosting, alongside Cloud Manager, Diagnose, Autoheal and Change Control.
- 01
Sign in as a supervisor, every 2 minutes
A failed sign-in stops a shift starting. The depot reverts to paper and the day is reconciled into Maximo afterwards, by somebody who was not there.
- 02
Create and save a work order, every 5 minutes
Emergent work goes unrecorded while the save step is broken. The compliance history for those assets has a gap that nobody can close later.
- 03
Submit an inbound integration message, every 5 minutes
Purchase orders, meter readings and finance postings queue silently, then arrive out of order. The failure is usually found in a month-end reconciliation.
- 04
Open the scheduling view for a crew, every 10 minutes
Dispatch falls back to a whiteboard. Crews are sent twice or not at all, and the schedule in Maximo stops matching the one being worked.
- 05
Sync a work order to a handheld, every 10 minutes
Field crews arrive without the job, the parts list or the permit. The travel is spent and the visit is repeated.
One night, two accounts of it
04:12. Every host is answering. Nobody can raise a work order.
An illustrative production event at the worst hour: after the batch window, before early shift. Both jobs are watching the same estate.
What the platform job saw
What the journey job saw
04:12
Every host answering, containers running, the database accepting sessions. No threshold crossed, so no alert.
Work order creation failed at the save step on production. Sign-in had passed twelve seconds earlier.
04:13
Nothing to report, correctly on its own terms.
The journey ran twice more and failed at the same step. A confirmed failure rather than a slow run.
04:14
Silent.
On-call alerted: the journey, the failing step, the environment, and the last successful run at 03:57.
06:05
The first signal of the morning is a supervisor on the phone, and the failure is already an hour and a half old.
The evidence bundle is with Diagnose. Sentinel stops at the handover.
Sentinel raises the alert and hands over evidence. The cause is Diagnose, and the pre-agreed repeat is Autoheal.
See what happens next →Boundaries
Three boundaries on Sentinel's coverage
Detection has edges, and they are set with you before the first check runs.
Detection ends at the alert and its evidence
Sentinel raises the alert and hands over the evidence. Nothing in it writes to your estate. Establishing the cause is Diagnose.
A check is only as good as the journey you define
Naming the journeys you cannot lose, agreeing the test account and the scoped test data, and stating the expected result for each is joint work before the first check.
Synthetic checks are samples
A five-minute cadence can miss a forty-second failure, and a scoped test asset will not see a fault limited to one site or one user's data. We state what your chosen interval leaves uncovered.
MaxIron Sentinel, frequently asked questions
- How is this different from infrastructure monitoring?
- Infrastructure monitoring samples hosts, containers, storage and certificates against thresholds. Sentinel completes the work a user does and times it. Keep both: Sentinel is the layer above, not a substitute for the one below.
- Does the work order check leave test data in production?
- No. A dedicated, non-privileged account works against a scoped test asset and location. Records are cleaned up or clearly marked so they never reach a real work queue, planner or report, and the scope is agreed in writing before the first check.
- How quickly does Sentinel notice a failure?
- Within the cadence of that journey plus one confirmation run. You set the cadence per journey, and we state what the interval you choose leaves uncovered.
- Where do alerts go?
- Wherever your operators already look: email, Teams, an ITSM queue or an on-call rota. Each alert carries the journey, the failing step, the environment and the last successful run.
- Does Sentinel work without Cloud Manager?
- Yes. It deploys standalone against classic Maximo 7.6.x and IBM MAS, and pairs with Cloud Manager, Diagnose and Autoheal where those are in place.
MaxIron products
MaxIron products that work alongside this
Tell us the last failure your monitoring missed.
One journey from your estate, the one that hurts when it stops. We show what Sentinel would have said, and at what time.
Bring this to the first call
- The incident record for the last Maximo issue a user reported before monitoring did
- The five journeys that must work before early shift
- Which environments matter out of hours, and which can wait until morning
- Who is on call, and which channel wakes them