The monitoring duty

Six duties kept running from the Tilbury control room of Csw Technologies Inc. Each one has a page in the client binder, a named owner and a written outcome. The scope is narrow on purpose.

What we hold

Watching is a shift job, not a dashboard licence

Our control room at 1 Queen St N, Tilbury - N0P 2L0, Canada (CA) keeps watch over the systems that our clients depend on. The work is arranged as six recurring duties, and the duties exist because each one answers a specific question that a monitoring dashboard cannot answer on its own. A dashboard will tell you that a disk is nearly full. It will not tell you who is on call at the moment it fills, which page in the binder covers it, or who agreed to the last change on that database.

We keep the watch small and deliberate. Each client gets a bay, a roster, a binder and a schedule. The bay watches, the roster names, the binder instructs and the schedule rehearses. Together they form the chain of custody for every alarm that crosses our wall, and the chain is what makes the record trustworthy when a client asks what happened on the night of the incident.

The six duties below are not menu items that a client buys in isolation. They are read together, the way a binder is read from the first tab to the last. A client who buys only a watch bay without a roster will eventually discover the gap at the worst possible hour. We say this plainly because it is the honest description of how monitoring fails.

C
Watch duty

Site Watch Bays

A site watch bay is a small, dedicated station inside the Tilbury control room assigned to one client portfolio. The bay carries a fixed set of checks: whether the public storefront resolves, whether the security certificate remains valid, whether the database disk is filling faster than the nightly cleanup can reclaim, and whether the scheduled backup job finished before the morning report was due. We treat those four checks as the vital signs of a small operator, because each one fails quietly and each one spoils a working morning.

Each bay is read aloud at the top of a shift. The engineer on duty walks the board, confirms the numbers, and writes a short note in the bay log. The note records what looked normal, what looked fragile and what will be watched more closely during the shift. A bay with no notes is treated as a bay that was not read, and that itself becomes a finding during review. The station layout keeps the wall behind the engineer and the binder in front, so the first move after an alarm is always toward the written page rather than toward the keyboard.

S
Roster duty

Alert Roster Runs

The alert roster is the daily check that the people who should answer an alarm can actually answer it. The run compares the live on-call calendar against the numbers written in the roster and against the pager assignments in the system. If a name has moved, a shift has been swapped or a contractor has left, the run finds it during daylight rather than at three in the morning. The roster is a living document, and the run is the habit that keeps it alive.

Every roster run ends with a signed line from the engineer who performed it and a count of the numbers confirmed. Where a number cannot be confirmed, the client contact is reached the same day and the correction is made in front of them. We do not accept a verbal promise that the number is fine, because the roster has to be true when nobody is available to explain it. The run also flags any client contact who has not confirmed availability in more than one cycle, since a silent rota is usually a rota in trouble.

W
Runbook duty

Incident Runbook Desks

An incident runbook desk is the physical and procedural station where a responder works through a known incident class. The desk holds a printed page with the first five decisions already written down: who to call, what to check first, what to switch off, what to preserve for the post incident review and when to escalate. The page is written in plain sentences and is rehearsed until the responder can scan it while a screen is flashing. Every page is short enough to fit on the desk surface next to the keyboard without obscuring the monitor wall.

Desks are assigned by incident class rather than by client, which means a responder who meets a certificate failure on one portfolio has already met it on another. That repetition is the whole point. Familiarity reduces the time between the alarm and the first correct action, and it reduces the chance that a responder will improvise a step that has never been tested. When a new failure appears that has no page, the page is drafted before the next shift and reviewed by a second engineer before it enters the binder.

T
Rehearsal duty

Pager Test Windows

Pager test windows are published intervals in which we fire the alerting chain on purpose and record exactly who answered and how quickly. The schedule is shared with the client at least a week in advance, and the test follows the same path a real alarm would take, from the monitoring check through the roster to the phone. A test that succeeds only because someone happened to be at a desk is not a success. We test during quiet hours and outside them, because the alarm does not choose a convenient time.

Every test produces a short record: the time the test was fired, the time the first acknowledgment arrived, the person who acknowledged and any stage of the chain that had to be bypassed. A silent pager is logged as a finding with an owner and a correction date. We keep the record even when the test passes, because a pattern of passes is the evidence a client needs when they are asked whether the alerting works. The windows also give the on-call engineer a low stakes moment to practise the phone tree before it matters.

E
Map duty

Escalation Map Reviews

The escalation map is a one page drawing of who calls whom, in what order and after how long. Every arrow on the map resolves to a phone that a real person answers within an agreed window, and every row carries a name rather than a role alone. Roles change and people move, so the map is reviewed on a fixed cadence and again whenever staffing changes. Review means redrawing, not patching. A map with three corrections scribbled in the margin is replaced by a clean sheet and a fresh signature.

During a review we walk the map from the first cell to the last and ask the same question at every step: if this number rings out, what happens next. If the honest answer is that nobody knows, the arrow is rebuilt. We also check that the map and the roster agree, because a disagreement between those two documents is the most common reason an escalation stalls in the middle of the night. The signed map lives in the binder and a copy is held with the client, and both copies are replaced at the same review.

O
Handover duty

On-call Handover Sittings

An on-call handover sitting is the structured meeting at the change of shift where the outgoing engineer passes the watch to the incoming engineer. The sitting is short but it is never skipped. The outgoing engineer reads the open alarms, the pending changes and the fragile systems aloud, and the incoming engineer reads the relevant binder page back before the sitting ends. Both engineers sign the handover line, and the signed line is the beginning of the next shift record.

We run handovers face to face inside the Tilbury control room whenever the engineers are in the building, and by a structured call when they are not. The call follows the same four headings in the same order, so a remote handover is not a looser version of the in person one. Anything that cannot be stated in words is treated as not handed over. If a client system is too complicated to describe in the sitting, that is a documentation defect and the binder is revised before the next change of shift.

How an engagement runs

From first conversation to signed handover

  1. Walk the system with the client

    We sit with the client and map what the system actually does, including the parts that were never documented. The output is a short list of conditions that would spoil a working morning.

  2. Draft the bay and the binder

    We choose the checks for the watch bay and write the first runbook pages. Each page is read back to a second engineer before it is accepted into the binder.

  3. Build the roster and the escalation map

    Names, numbers and response windows are agreed in writing. Every entry is confirmed during daylight before the first shift runs under the new roster.

  4. Rehearse the pager chain

    We fire the chain on purpose inside a published test window and record who answered. Findings are corrected before the first live watch night.

  5. Run the first monitored shift

    The bay is watched by a named engineer and the bay log is kept in full. The client sees the first log before the first invoice is issued.

  6. Review and hand the binder over

    At the end of the first cycle we review the map, the roster and the pages with the client. The binder belongs to the client from that day forward.

Ask for a watch bay assessment

Send a short note describing the system that keeps you awake. We will reply with an honest view of whether our control room is the right place for it. Reach us at alert@cswtech.lat or on +19285925347.

Contact Csw Technologies Inc. →