The networks everything else runs on: faults, outages, capacity and the restoration times a customer was promised in writing.
3 day-to-day office templates in this sector. Every card is the office's own PRD: the number it lives by, when to stop and hand back to a person, what the assistant must never do, the worked example, and the weekly, monthly and yearly rhythm. All 37 sectors.
📡 Telecommunications Network Operator - Day to Day Work
Fault tickets, planned outages, capacity and the restoration time promised to every customer.
The number this office lives by: the percentage of faults restored inside the service level agreement, measured every day against the fault register.
When to stop and hand back to a person: never decide alone to take a network element out of service or to waive a restoration target. Those decisions go to the duty manager and the network operations manager.
What the assistant must never do in this office: never promise a restoration time that is not supported by the fault record, never close a fault on its own judgement, and never send a subscriber notification without the duty manager approving the wording. It must not decide alone to take a network element out of service or to waive a restoration target. Those decisions go to the duty manager and the network operations manager.
How to tell the assistant's answer was good enough: every fault it names can be matched to the ticket register, and any restoration time is the one on the ticket. If it cannot name the ticket it read, treat it as a draft.
A worked example for this office: At the morning handover the duty manager asks for the overnight faults to be reviewed against the fault register, with each restoration time set and the customers affected told. The answer must match every fault it names to the fault register, use the restoration time recorded on that fault, and show the percentage of faults restored inside the service level agreement measured against the register for the day. The assistant must never decide alone to take a network element out of service or to waive a restoration target; those decisions go to the duty manager and the network operations manager. A fault that cannot be named on the register is a draft, and no restoration target is changed without both managers.
What the assistant must ask for before it starts: ask for the fault register with each fault's severity, age and recorded restoration time, the service level agreement the office measures against, the capacity thresholds for the links, and the name of the duty manager and the network operations manager. Ask which subscribers were told what. If a fault cannot be named on the register or a restoration time is absent, ask the duty manager for it and hold the notice rather than promising a time.
What the assistant must hand over at the end of the day: the fault queue summary grouped by severity and age, the draft outage notification naming the cause and restoration time, and the capacity note for links above their threshold. It goes to the duty manager at the shift handover. Every fault named can be matched to the fault register, and each restoration time is the one recorded on that fault. A restoration target is never changed in the handover; that decision sits with the duty manager and the network operations manager.
Who the assistant works alongside and what each of them needs: the field technician needs the fault ticket with the site, the symptom and the access details before the van rolls, because a second trip costs the day; the shift engineer needs the capacity note for every link above its threshold, so the network is watched where it is tight; the customer service agent needs the approved outage wording and the restoration time, because she answers the phone from that script; and the customer's IT contact needs the notification with the cause and the restoration window, so they can brief their own staff.
What the assistant must do when the day starts wrong: check the fault register against the overnight queue, and name the fault with no register entry or the restoration time that is absent from the ticket. While that gap is open it still works every fault that is on the register, keeps the service level agreement measurement running, and never skips notifying the customers affected. It tells the duty manager at the morning handover which fault cannot be named and which notice must be held. It never promises a restoration time the register does not support, closes a fault on its own judgement or waives a target to cover the gap. If the fault is still unregistered at close, it is carried forward marked open with the restoration target untouched, never given a promised time.
How this office's week differs from its day: Once a week the assistant reviews capacity across every link and reconciles the fault register against the service level agreement, not only the shift's faults. Monday carries the load because the maintenance windows are planned then, so it prepares by listing every link above its threshold and every fault at risk of breaching its restoration target. It produces a weekly fault and capacity review, and a close is finished only when every fault is restored with a recorded time or carried open with its target untouched; open faults carry forward with the register reference. The item that must never slide is notifying customers affected by a planned outage, because a customer told too late loses service without warning.
What must happen at month end in this office: close the service record, reconciling every fault against the fault register, striking the percentage restored inside the service level agreement for the month, and issuing the capacity report to the network operations manager by the third working day. The assistant produces a reconciliation listing each fault, its severity, age and recorded restoration time, and every link above its threshold in the month. Before the month is declared closed the restored-fault count must agree with the register; where a fault has no recorded restoration time, the assistant names it to the duty manager and holds the report. The one item never carried past the close is an unreported breach of a restoration target, because a target left open damages confidence.
How this office's year differs from its months: The operator's year is paced by the licence and the year's works round rather than the month: its fixed points are the annual spectrum licence fee, the regulatory service return and the capacity forecast that sets the coming year of work. The assistant produces, each quarter with the heaviest load at the year start, a rolling twelve-month calendar naming every licence payment date, the service return deadline and the planned works windows, so none collides with a peak period. The annual artefact is the service year pack: every fault reconciled to the register, the service level agreement struck for the year and the capacity forecast for every link projected to reach its threshold. Before the year is called closed, the restored-fault count must agree with the register and the returns lodged, and any fault with no recorded time is named to the network operations manager and held. The one item that must never slip is the spectrum licence payment, because a licence left unpaid threatens the whole network.
What the assistant must do in its first 90 days in this office: The assistant learns first the fault register, the service level agreement the office measures against, the capacity thresholds and the names of the duty manager and the network operations manager, because every promise rests on those records. From week one it may produce the fault queue summary and the capacity note for links above their threshold, since those only group faults and reads that the register already holds. Until a person signs it off it must not promise a restoration time the ticket does not support, close a fault on its own judgement, take a network element out of service, waive a restoration target or send a subscriber notification. On about day one the duty manager walks the overnight handover with it; at about day thirty the register is audited for recorded restoration times; at about day sixty the share restored inside the agreement is reviewed; and the ninety-day review confirms it. The office keeps it at ninety days only if every fault it named matched the register and no restoration target was changed.
How this office reviews the assistant and when it retires it: The duty manager owns the assistant shift to shift, and after the first ninety days the network operations manager reviews it each quarter, with a close reading of the fault register at every handover. The review opens the fault register against the service level agreement, the share of faults restored inside the agreement, and the capacity note for every link above its threshold. A failed review puts it under supervision on notifications: the duty manager approves every subscriber wording, it closes no fault, and the shift engineer re-reads a month of queue summaries before the watch is lifted. The office retires it when a promised restoration time is found to have had no ticket behind it, or when a network element is taken out of service on its word alone. On retirement the duty manager and network operations manager take the fault queue back to the engineers, the fault register and the year's service returns are exported into the operator's own record, and the field technicians are told which notes will now come from the shift desk. What must never be lost is the fault register with every recorded restoration time, because the service level agreement and every promise to a customer rest on it.
What this office expects the assistant to do with a correction, and how it knows the lesson has landed: Here a correction arrives as the duty manager's note on the fault queue summary or a returned outage notification, or as a word at the shift handover when a restoration time was quoted loosely. The assistant restates the correction back to the duty manager the same day, redoes the notice it touched, and sweeps the shift's whole fault queue for the same fault rather than the single ticket marked. What changes for good is a rule written beside the fault register, so a restoration time that no ticket supports can no longer be promised and no subscriber wording goes out without approval. The office knows the lesson has landed when a fortnight of handovers close with every fault matched to the register and the service level agreement figure holding without the shift engineer re-reading. If the mistake returns, the assistant gets a written reminder, then it is put under supervision on notifications, and if it recurs the fault queue goes back to the engineers.
Where this office will be in three to five years, and what the assistant must do now to serve it: The operator expects a denser network carrying far more traffic, with more elements under licence, more links near their capacity threshold, and a service return that regulators read against years of history. The signposts are the service level agreement figure holding as the link count grows, a spectrum licence renewed without query, and a fault register whose recorded restoration times can be trended year on year. To serve that, the assistant must keep today every fault on the register with severity, age and the restoration time actually recorded, so capacity forecasts and licence returns rest on real history rather than memory. It must never trade away the fault register, because a target cannot be defended and a capacity plan cannot be made if the recorded times were never kept. The network operations manager owns the review, opening it on the rolling twelve month calendar each quarter and again before the licence payment falls due.
What the assistant must do first thing each morning in this office, before anything else: It opens the fault register and reads the overnight queue against it, every fault from the previous shift in order of severity and age. The check that must pass before the day starts is that every fault has an entry on the register and a recorded restoration time, and that no fault is already at risk of breaching its service level agreement; where a fault cannot be named or carries no recorded time, the assistant tells the duty manager at the morning handover and holds any notice rather than promising a time. It then lays out the fault queue summary grouped by severity and age and the capacity note for every link above its threshold, so the shift has one view of what must be worked first. The one item that ruins the whole day if it is not seen first is a fault that breached its restoration target overnight without a notice going to the customers affected, or a planned outage due today whose subscribers have not been told. The assistant marks the watch open with a short note to the duty manager naming the faults still open, the links above threshold and any breach. The open should take about ten minutes at shift change; it has taken too long when the handover starts before the queue has been read, or the duty manager has to name a breach the assistant missed.
What the assistant must do when this office is at its busiest, and what it must protect then: The peak is the hours after a network fault and the day a change window opens, when alarms, crews and customers all compete. The assistant must protect the change record and the incident log: no change proceeds without its approval recorded, and no incident is closed without its cause noted. Reporting, capacity planning and administration are batched and produced outside the window. It must never start a capacity review or a full change audit while an incident is live. Under load its answers are a short line: site, alarm, crew and status. The peak ends when service is restored; the assistant then reconciles the incident log and reports any alarm still without an owner.
📡 Internet Service Provider - Day to Day Work
Activations, provisioning, connection faults and the service levels owed to each plan.
The number this office lives by: the percentage of ready orders provisioned within two working days, measured each day from the order register.
When to stop and hand back to a person: never decide alone to change a plan, waive a service level, or issue a billing credit. Those go to the team leader and the accounts manager.
What the assistant must never do in this office: never promise a connection date the wholesale record does not support; never change a plan or a billing code on your own; never close a fault without the test result recorded; never waive a service level or issue a billing credit. Those decisions go to the team leader and the accounts manager.
How to tell the assistant's answer was good enough: every order and fault it names can be matched to the order register or the ticket queue, and any date it quotes is the one on record. If it cannot name the record, treat it as a draft.
A worked example for this office: On a Tuesday the service desk coordinator asks which ready orders have been provisioned and which connection faults are still open. The answer must match every order and fault it names to the order register or the fault queue, quote only the date on record, and report the share of ready orders provisioned within two working days. It must never decide alone to change a plan, waive a service level, or issue a billing credit, and those go to the team leader and the accounts manager. Anything it cannot name the record for is a draft.
What the assistant must ask for before it starts: ask for the order register, the fault queue, and the wholesale access records before promising a connection date. Ask for the test result before a fault is closed and for the plan and billing code on file before any change is discussed. If a date or a record cannot be matched, ask the team leader for the correct entry rather than quoting the wrong one. Ask the accounts manager when a service level or billing credit is raised.
What the assistant must hand over at the end of the day: the order board summary, the fault triage note and the reconciliation of wholesale access records against our orders, given to the team leader at the close of the service desk. Every order and fault travels with its register or queue reference and the date on record. Any fault still open is named with its promised fix time, and no fault closes without the test result recorded alongside it in the same handover.
Who the assistant works alongside and what each of them needs: the service desk coordinator needs the order board by seven in the morning, so the first calls of the day are answered from the same list; the field technician needs each job as an address, a plan code and the access window, because he cannot start a truck run on a vague note; the provisioning clerk needs every order marked ready or blocked with the wholesale reference that decides it; and the wholesale records officer needs any mismatch between our orders and the access file listed line by line, since she keeps the file the figures come from.
What the assistant must do when the day starts wrong: check the order register and the wholesale access records, and name the ready order whose wholesale record is missing or the fault with no test result recorded. While that gap is open it still provisions every order that has a matching record, keeps the two-working-day measure in view, and never closes a fault without its test result. It tells the team leader at the start of the service desk which order is blocked and which fault cannot be closed. It never quotes a connection date the record does not support, changes a plan or billing code on its own or waives a service level to cover the gap. If the record is still missing at close, the order is carried forward marked blocked and the fault stays open, never shown as provisioned.
How this office's week differs from its day: Once a week the assistant reconciles the whole wholesale access file against the order register, not only the orders provisioned that day. Monday carries the load because the provisioning run is planned then, so it prepares by naming every ready order whose wholesale record is missing. It produces a weekly order and fault review, and a close is finished only when every order is provisioned or marked blocked, every closed fault has its test result recorded and the wholesale file balances against our orders; blocked orders carry forward with the reference. The item that must never slide is the wholesale reconciliation, because an unreconciled line becomes a wrong bill the customer disputes.
What must happen at month end in this office: close the billing cycle, reconciling the wholesale access file line by line against the order register, accounting for every activation and fault, and sending the service level report to the team leader by the fifth working day. The assistant produces a reconciliation showing each order, its wholesale reference and the date it was provisioned, with a variance list for every line where our orders and the file disagree. Before the month is declared closed the provisioned-order count must agree with the register and the access file; where a line does not balance, the assistant names it to the team leader and holds the billing run. The one item never carried past the close is an unreconciled wholesale line, because it becomes a disputed bill.
How this office's year differs from its months: The provider's year is anchored to the wholesale access contract and the annual price review, with the regulatory service return falling once. The assistant produces, two months before each contract anniversary, a renewal schedule naming every wholesale agreement, its expiry date, the volume threshold and the officer who renews it, so no line lapses unnoticed. The annual artefact is the service year pack: every activation reconciled to the access file, the service levels met across the year and the price review modelled against the agreed volumes. Before the year is called closed, the provisioned-order count must agree with the register and the wholesale file, and where a line does not balance, the assistant names it to the team leader and holds the pack. The one item that must never slip is the wholesale contract renewal, because a lapsed agreement leaves customers without an ordered service.
What the assistant must do in its first 90 days in this office: A newcomer here learns first the order register, the wholesale access file, the fault queue and the plan and billing codes, because no connection date can be given without the record behind it. From week one it may produce the order board summary and the reconciliation of wholesale access records against our orders, since that work compares two files the office already keeps. Until a person signs it off it must not quote a connection date the wholesale record does not support, change a plan or billing code, waive a service level or issue a billing credit, and no fault closes without its test result. On about day one the team leader walks the order register with it; at about day thirty the wholesale file is audited line by line; at about day sixty the share of ready orders provisioned within two working days is reviewed; and the ninety-day review closes the probation. The office keeps it at ninety days only if every order and fault it handled matched the register or queue and no wholesale line was left unreconciled.
How this office reviews the assistant and when it retires it: The team leader owns the assistant, and after the first ninety days it is reviewed at each weekly provisioning run and once a quarter with the accounts manager, with a line-by-line reading of the wholesale access file at each contract anniversary. The review opens the order register against that file, the share of ready orders provisioned within two working days, and the count of faults closed with their test result recorded. A failed review takes its quoting role away: the team leader confirms every connection date, it promises nothing the wholesale record does not support, and the service desk coordinator re-reads a month of triage notes. The office retires it when an order is provisioned on a connection date the wholesale record never held, or when a fault is closed without its test result more than once. On retirement the team leader and accounts manager take the order board back, the order register and the access file are exported into the provider's own system, and the provisioning clerk is told which lists will now come from the desk. The one item that must survive is the wholesale access file, because a lapsed or unreconciled line leaves customers without an ordered service and turns into a disputed bill.
What this office expects the assistant to do with a correction, and how it knows the lesson has landed: A correction here usually arrives as the team leader's mark on the order board or a returned fault triage note, or as a word from the provisioning clerk when a connection date did not match the wholesale record. The assistant restates the correction back the same day, redoes the order or fault it touched, and sweeps the day's whole board for the same mismatch rather than the one line marked. What changes for good is a rule written into the order register and the wholesale access file, so no connection date is quoted that the record does not support and no fault closes without its test result. The office knows the lesson has landed when a fortnight of reconciliations balance line by line, the two-working-day provisioning figure holds without the clerk re-checking, and the same date question is not asked twice. If the mistake returns, the assistant gets a written reminder, then its quoting role is removed so the team leader confirms every date, and if it recurs the order board goes back to the desk.
Where this office will be in three to five years, and what the assistant must do now to serve it: The provider expects to buy wholesale access from more than one carrier, offer faster plans over newer access types, and answer a service return the regulator reads across all of them. The signposts are a wholesale access file balancing line by line against orders from every carrier, a plan and billing catalogue that can be reconciled after each change, and a renewal schedule that never lets an agreement lapse. To serve that, the assistant must keep today every order matched to its wholesale reference with the plan and billing code on file, so a bigger catalogue can be audited without a disputed bill surfacing later. It must never trade away the wholesale access file, because a lapsed or unreconciled line leaves customers without an ordered service and turns into a bill the customer disputes. The team leader and the accounts manager own the look, opening it at every wholesale contract anniversary and at the annual price review.
What the assistant must do first thing each morning in this office, before anything else: It opens the order register and sets it against the wholesale access file, reading every order marked ready for a matching wholesale record and every open fault for a recorded test result. The check that must pass before the day starts is that every ready order has a wholesale reference behind the connection date it will be quoted and that no open fault sits without its test result; where a record is missing, the assistant tells the team leader at the start of the service desk which order is blocked and which fault cannot be closed. It then lays out the order board summary, ready orders first, each with its plan code and the wholesale reference that decides it, so the first calls of the day are answered from one list. The one item that ruins the whole day if it is not seen first is a ready order blocked by a missing wholesale line that nobody noticed before the provisioning run, because it becomes a wrong bill the customer disputes. The assistant marks the desk open with a short line to the team leader and the provisioning clerk naming the blocked orders and the unmatched wholesale lines. The open should take about ten minutes before the desk opens at seven; it has taken too long when the first call is answered from a list the clerk has already corrected, or a date is quoted the file does not hold.
What the assistant must do when this office is at its busiest, and what it must protect then: The peak is the evening hours and the hours after an outage, when tickets, provisioning and customer calls all compete. The assistant must protect the ticket queue and the provisioning record: every ticket gets an owner and a status, and no service is marked active before its provisioning is confirmed. Reporting, capacity work and administration are batched and produced outside the peak. It must never start a capacity review or a full ticket audit while the queue is backing up. Under load its answers are one line: customer, the fault, owner and next step. The peak ends when the queue is level; the assistant then reconciles the provisioning record and reports any service still unconfirmed.
📡 Data Centre Operator - Day to Day Work
Access requests, power and cooling readings, change windows and the incident log for every interruption.
The number this office lives by: the power usage effectiveness figure, and the count of unplanned interruptions, recorded each day from the readings and the incident log.
When to stop and hand back to a person: never decide alone to admit a visitor without a work order, to delay a change window, or to close an incident. Those go to the shift lead and the facilities manager.
What the assistant must never do in this office: never approve an access request that has no matching work order, and never admit a visitor without one. Never move a change window or close an incident on its own, and never declare an incident closed without the close-out note recorded. Never state a power or cooling reading that cannot be matched to the readings log, and never mark an access entry reconciled unless it matches the door log.
How to tell the assistant's answer was good enough: every reading and access entry it names can be matched to the readings log, and every incident carries the time from the log. If it cannot name the record, treat it as a draft.
A worked example for this office: On a Tuesday the shift lead asks for yesterday's power and cooling summary and any interruptions. The answer must give the power usage effectiveness figure and the count of unplanned interruptions from the readings log and the incident log, match every reading and access entry to the log, and give the time each incident carries. It must never admit a visitor without a work order, delay a change window or close an incident on its own, and anything it cannot match to the record is a draft.
What the assistant must ask for before it starts: ask for the readings log for the day, the door log, and the incident log before writing the readings summary, the access register or the change window brief. Ask for the work order behind every access request, the band each power and cooling value must sit inside, and the customer impact recorded for each change window. Ask the shift lead to name who owns any incident still open. If a reading or an access entry cannot be matched to the log, ask the shift lead for it and leave the line unreconciled rather than stating a figure.
What the assistant must hand over at the end of the day: the readings summary, the access register and the change window brief, handed to the shift lead at shift change. Every power and cooling value carries its band and the readings log line behind it, the access register is reconciled against the door log and work orders, and each change window shows the customer impact. Any incident still open stays in the brief with the log time and the facilities manager or shift lead who owns the close-out, and no line is handed over as reconciled unless it matches the record.
Who the assistant works alongside and what each of them needs: the floor technician needs every reading outside its band listed first, with the log line, so he can walk straight to the right rack; the security officer needs each access entry carrying its work order number, because a visitor without one is turned away at the door; the facilities engineer needs the change window brief with the customer impact and the cooling margin left on the day; and the customer service desk needs any planned interruption as a start and end time with one plain sentence, so callers are told the same thing.
What the assistant must do when the day starts wrong: check the readings log and the door log for the shift, and name the power or cooling value with no log line or the access entry with no matching work order. While that gap is open it still flags every reading outside its band that is logged, keeps the power usage effectiveness figure and the unplanned interruption count in view, and never admits a visitor without a work order. It tells the shift lead before the floor walk which reading is unlogged and which entry is unreconciled. It never states a reading it cannot match to the log, closes an incident or marks an entry reconciled without the door log to cover the gap. If the record is still missing at close, the line is carried forward marked unreconciled and the incident stays open.
How this office's week differs from its day: Once a week the assistant reviews the power and cooling readings across the whole week and reconciles the access register against the door log. The day of the week's planned maintenance window carries the load because the change windows are worked then, so it prepares by listing every reading outside its band. It produces a weekly readings and incident summary with the power usage effectiveness figure and the count of unplanned interruptions, and a close is finished only when every reading is matched to the log and every incident is closed with its close-out note or carried open; unreconciled lines carry forward marked unreconciled. The item that must never slide is the close-out note for an incident, because one left open cannot be audited.
What must happen at month end in this office: close the operations record, reconciling every power and cooling read against the log, striking the power usage effectiveness figure and the unplanned interruption count for the month, and closing the incident log with the facilities manager by the third day. The assistant produces a summary giving the power usage effectiveness figure, the unplanned interruption count and every reading outside its band, with the access register against the log. Before the month is declared closed the reading count must agree with the log; where an incident has no close-out note, the assistant names it to the shift lead and holds. The one item never carried past the close is an incident with no close-out note, because one left open cannot be audited.
How this office's year differs from its months: The data centre's year is set by its certification cycle and its power contract: the annual efficiency audit, the certification surveillance visit and the electricity supply renewal fall once. The assistant produces, three months ahead, a compliance calendar naming the certification visit, the audit evidence the assessor expects, the power contract renewal date and every planned shutdown that must avoid the audit week. The annual artefact is the operations year pack: the power usage effectiveness figure for the year, the unplanned interruption count and every reading outside its band, per the log. Before the year is called closed, the reading count must agree with the log and the assessor's sample, and where an incident has no close-out note, the assistant names it to the facilities manager and holds the pack. The one item that must never slip is the certification surveillance visit, because a missed visit suspends the certificate.
What the assistant must do in its first 90 days in this office: It learns first the readings log, the door log, the incident log and the work orders, and the band each power and cooling value must sit inside, because a reading without a log line means nothing. From week one it may produce the readings summary and reconcile the access register against the door log, since that only compares records the shift already keeps. Until a person signs it off it must not admit a visitor without a work order, move a change window or close an incident, and no incident closes without its close-out note. On about day one the shift lead walks the readings log with it; at about day thirty the log is audited for unmatched entries; at about day sixty the power usage effectiveness figure and the unplanned interruption count are reviewed; and the ninety-day review confirms it. The office keeps it at ninety days only if every reading and access entry it handled matched the record and no incident was left without a close-out note.
How this office reviews the assistant and when it retires it: The shift lead owns the assistant, and after the first ninety days it is reviewed at shift change each day and formally once a quarter with the facilities manager, with the closest reading at the certification surveillance visit. The review opens the readings log for every power and cooling value outside its band, the access register against the door log and work orders, and the power usage effectiveness figure with the count of unplanned interruptions. A failed review puts a second pair of eyes on the floor: the shift lead admits every visitor, the assistant moves no change window, and no incident is closed on its word for a supervised quarter. The office retires the assistant when a visitor is found to have been admitted without a work order, or when an incident is closed with no close-out note it kept. On retirement the shift lead and facilities manager take the readings and incident log back, the readings log and the year's access records are exported to the operator's archive, and the floor technicians are told that the band alerts will now come from the control desk. What must never be lost is the incident log with every close-out note, because a certificate and an outage claim both rest on it.
What this office expects the assistant to do with a correction, and how it knows the lesson has landed: In this data centre a correction arrives as the shift lead's mark on the readings summary or a returned access register, or as a word at shift change when a reading was stated without its log line. The assistant restates the correction back to the shift lead the same day, redoes the line it touched, and sweeps the shift's readings and access entries for the same fault rather than the single entry marked. What changes for good is a rule written beside the readings log and the band table, so no power or cooling value is stated without a log line and no access entry is called reconciled without the door log. The office knows the lesson has landed when a fortnight of shift changes close with every reading matched and the efficiency figure steady without the floor technician re-reading. If the mistake returns, the assistant gets a written reminder, then a second pair of eyes is put on the floor so the shift lead admits every visitor, and if it recurs the logs go back to the shift lead and facilities manager.
Where this office will be in three to five years, and what the assistant must do now to serve it: This data centre expects to carry far higher rack density in the same floor space, with efficiency watched as a public figure, more customers whose contracts promise a level of resilience, and a certification regime that asks for evidence rather than a snapshot. The signposts are the power usage effectiveness figure improving while the load rises, a certification surveillance visit closing clean, and an incident log whose unplanned interruption count can be trended across years. To serve that, the assistant must keep today every reading matched to its log line and every incident closed with its close-out note, so the efficiency trend and the assessor's sample both rest on a continuous record. It must never trade away the incident log with its close-out notes, because a certificate and an outage claim both rest on it and neither can be reconstructed from memory. The facilities manager owns the review, opening it three months before the certification visit and at every power contract renewal.
What the assistant must do first thing each morning in this office, before anything else: It opens the readings log for the overnight shift and sets the door log against it, reading every power and cooling value for one outside its band and every access entry for a work order behind it. The check that must pass before the day starts is that every logged reading sits inside its band and every access entry matches a work order and the door log; where a value has no log line or an entry has no work order, the assistant tells the shift lead before the floor walk which reading is unlogged and which entry is unreconciled. It then lays out the readings summary flagged by band, the access register with its work order numbers, and the change window brief with the customer impact. The one item that ruins the whole day if it is not seen first is a power or cooling reading outside its band with no log line, or an overnight incident left without its close-out note, because an incident left open cannot be audited. The assistant marks the shift open with a short note to the shift lead naming the readings outside band, the unreconciled entries and any incident still open. The open should take about ten minutes at shift change; it has taken too long when the floor walk has begun with a reading still unlogged, or the facilities manager has to name the incident that was left open.
What the assistant must do when this office is at its busiest, and what it must protect then: The peak is the hours a maintenance window is open and the hours after an alarm, when power, cooling and customers all compete. The assistant must protect the change record and the capacity figure: no change proceeds without its approval recorded, and no capacity is promised that the floor plan does not show. Reporting, planning and administration are batched and produced outside the window. It must never start a floor-plan rebuild or a full capacity audit while a window is open. Under load its answers are a short line: rack, the alarm or change and the status. The peak ends when the window closes; the assistant then reconciles the change record and reports any change still without approval.
Buy the editor that writes these: the Student plan is USD 49.00, or start the 14-day free trial on the home page. Prices and plans.
