Your first five minutes
- Pick an example. Start with rack deployment, hardware replacement, a planned network change, or Other / custom process. The custom option starts with five generic steps and illustrative estimates to replace. Selecting a new example replaces the current unsaved inputs.
- Make it yours. Give the process a name. In speed or risk mode, edit the five steps to match your site. Keep the sequence simple.
- Choose a question. Speed asks “When are we ready?” Risk asks “How often do we get through cleanly?” Staffing asks “Can our team cover the work?”
- Check operating assumptions. Review the change window, corrective delay, or staffing reserve. Then click Explore the possibilities.
- Try one improvement. Choose a step and reduction, then click Compare. Discuss which assumptions need real evidence.
At the booth: ask the visitor which operational question matters most, explore one result, and compare one improvement. Save the link if they want to return later. Use Start fresh before the next visitor to clear the current inputs.
The sample values are invented demonstration inputs, not data center benchmarks. The initial chart is already calculated from the displayed example. After editing inputs, a notice tells you to run again. Switching analysis tabs clears all results and comparisons immediately; click Explore the possibilities to calculate the selected analysis. Your inputs stay available across tabs.
Choose an example or your own process
- Rack deployment
- Plan and approve → rack and stack → power and cable → configure and test → handover. Discuss readiness dependencies and acceptance evidence. Power and cooling availability are prerequisites to verify, not quantities this tool calculates.
- Hardware replacement
- Triage and scope → approve change → replace hardware → validate service → close and hand over. Discuss preparation outside the execution window and work needed to establish service readiness.
- Planned network change
- Review change and MOP → confirm readiness → execute change → validate service → handover and close. MOP means method of procedure: your approved sequence for carrying out a change. The example supplies workflow labels, not equipment instructions.
- Other / custom process
- Give your process a name, rename Step 1 through Step 5, and enter your own times, issue chances, and operating assumptions. Starter numbers are placeholders, not estimates of your process. Results stay blank until you run.
In staffing mode, estimate an entire work order directly. The five-step timings remain available in the other modes but do not determine technician effort.
Use the right kind of hours
Elapsed hours are time on the clock. Technician-hours add up each person's effort. Two technicians working for three hours use six technician-hours, even though only three hours pass.
| Input | Plain-English meaning |
|---|---|
| Normal step hours | Average elapsed duration when there is no extra corrective intervention. Include normal waiting if relevant. Exclude the corrective delay entered separately. |
| Issue chance | Out of 100 executions of that step, how many need correction or fail a check? This is not the probability of an outage. |
| Timing variation | How much normal duration or effort changes between jobs. Low, moderate, and high mean standard deviations of 15%, 35%, and 65% of the average. Durations are always positive and can have a long tail. |
| Extra time to fix a problem at a step | Also called corrective delay. Enter 2 hours and each affected step adds two hours to the clock; two affected steps add four hours. Each step can have at most one issue per work order. Two issues add two delays. This is not labor effort and is not added to the staffing model. |
| Ready-for-service target | The target for the entire five-step process, including normal waiting and corrective delay. |
| First step inside the window | The point at which execution begins. This step and all later steps count against the change window. Earlier preparation is excluded. |
| Window and rollback reserve | The approved execution period and time held back. A six-hour window with one hour held back leaves five hours of usable execution time. The tool does not simulate an actual rollback. |
| Technician-hours per order | Total effort from everyone working on one order. Include expected corrective labor here if appropriate; it is separate from elapsed corrective delay. |
| Work orders per week | Typical weekly volume. Demand variation creates different weekly volumes around that input. The moderate and high settings use 25% and 50% relative variation. |
| Available technicians | The pool you can allocate to this work, not necessarily every employee at the facility. |
| Hours per technician | Weekly productive hours before the additional protected reserve. Account for leave, breaks, and training consistently. |
| Protected reserve | Share of those hours held for other duties, such as response work or escorts. Do not deduct the same duties twice. At 32 hours and a 20% reserve, 25.6 hours remain per technician. |
A range, not a promise
Each run explores 5,000 possible work orders or weeks. The animation is a short visual transition. The calculation uses the same random sequence each time, so changing an input has a repeatable effect.
Speed and change-window readiness
The average adds all simulated completion times together and divides the total by 5,000. P90, or “9 in 10,” is the time within which 90% of modeled runs finish. It is not a guarantee.
The main chart includes the entire workflow and corrective delays. The operator readout separately measures execution from your chosen window-start step. Miss usable change window means this execution time exceeds window length minus rollback reserve. It does not mean downtime occurred or tell an operator to continue past an approved limit.
Risk and first-pass completion
First-pass yield is the share of orders with no issue at any of the five steps. The main cards are simulated estimates. The operator card is an exact calculation from the entered independent probabilities, so small differences are normal.
The sensitivity table asks what would happen if each step's current issue probability fell by 30%, one step at a time. Gains are percentage points: 80% to 83% is a gain of 3 points. These separate benefits should not be added together. “Delay per 100 orders” is that step's expected corrective elapsed time in the displayed scenario.
Staffing and workload coverage
The model varies weekly order counts and each order's effort, then divides total technician-hours by the usable hours per technician. P90 team needed rounds up to whole technicians. This is a workload equivalent, not a roster or a minimum safe staffing recommendation.
Mean uncovered work averages hours above the entered team's capacity across all weeks, including weeks with zero excess. P90 capacity gap compares the rounded P90 need with your entered team. The chart does not carry an unfinished backlog into later weeks.
Reading the chart
The horizontal axis shows hours or technician equivalents. Taller bars mean more simulated outcomes fall in that range. The lighter bars show the baseline; the darker bars show the comparison. The dashed line marks the target or available team. The risk chart instead shows the entered issue chance for each step.
Change one assumption at a time
In speed mode, the slider reduces one step's normal duration. In risk mode, it reduces one step's issue chance. A 30% reduction takes an 8% issue chance to 5.6%, not to zero. In staffing mode, it reduces total effort per work order.
The upper metrics and operator readout show the improved scenario after comparison; the light bars retain the baseline. The slider does not change results until you select Compare. A new normal run replaces the baseline and clears the comparison. A risk improvement can reduce corrective delay, while a speed improvement leaves the issue probabilities unchanged.
Use the result to choose what to investigate, such as readiness checks, handover evidence, or avoidable waiting. The tool does not establish that a proposed intervention will achieve the chosen percentage.
Keep the conversation going
Select Take this conversation with you, agree to save the estimates, and create a session link. Copy the link to keep it or share it. A retrieval code and optional QR code also restore your inputs, operating assumptions, mode, and applied comparison. No email address or marketing enrollment is required.
Anyone with the link or code can view or delete the saved session. Use nonconfidential estimates. Saved sessions expire after 30 days. Open a saved link to find its deletion option. The manual opens in the same tab. Use Return to simulator to resume your working scenario; a temporary draft stays in this browser tab.
A phone must be able to reach the demo's server. A localhost link works only on the computer running it. For an event or later access, your booth associate needs a reachable network or hosted address.
If results look old, click Explore the possibilities. If a validation message appears, correct the inputs named in the message—for example, rollback reserve must be shorter than the change window. Older saved sessions load with the original zero-correction, fixed-demand assumptions; review them before adding the deeper analysis.
What this model leaves out
- Issues and step durations are independent. Common causes, such as a readiness problem affecting several steps, are not modeled.
- There is one fixed corrective delay per issue, no repeat failures, no early abort, and no explicit rework loop. All five steps are counted.
- Staffing uses a flexible pool with independent effort per order. Weekly demand varies, but skills, shift timing, two-person tasks, queueing, and backlog carryover are not modeled.
- The tool does not calculate electrical load, cooling adequacy, redundancy, service availability, Tier compliance, outage probability, or financial ROI.
Use approved site procedures and engineering review for operational decisions. This simulator supports a planning conversation; it does not authorize or prescribe a live change.
Bring the records behind the work
For a stronger follow-up, gather a representative set of completed work orders, normal and corrective timestamps, failed checks, change-window overruns, weekly arrivals, and technician effort by role. Keep elapsed time and labor effort separate. Include difficult jobs as well as clean completions, and identify shared disruptions.
Start with one consistent job type and one site or team. Compare the real distribution with the simulator's assumptions before using results to size an intervention.
Why these operating themes?
Uptime Institute's management and operations criteria cover staffing, maintenance, training, planning, and operating conditions. Its articles on methods of procedure and data center staffing provide context for discussing procedural work and team capacity.
These references inform the questions to ask. They do not supply the demo's numbers, validate its mathematics, or imply endorsement. All presets and improvement percentages are illustrative.