# Scenarios > Ask "what if?" of a map, such as Black Friday or a product launch, play it over time, see which part runs short first, and apply the fix. Your orders API copes fine today. Will it on Black Friday, when seven times as many people show up at midnight? A scenario plays that day on your map and shows you the answer before your customers find it: which parts run short or fail, when, for how long, and what would fix them. A scenario can be a busy day, a TV ad that sends everyone at once, a launch, steady growth, or something breaking. Flozu has 9 ready-made ones, and you can make your own. ## Try it On the **Checkout system** template: 1. Open **Scenarios** in the map’s menu and pick **Black Friday**. A timeline appears along the bottom, and the map shows the busiest moment, ten past midnight. 2. Read the card at the top of **What happens**: 2 of 3 parts need room, the busiest at 438%. 3. On the canvas, the Email worker is outlined in red. About 88 emails a second reach it, and it can send 20. 4. Under **Fix it**, Flozu says to give it 126 a second, or run 7 copies. Select **Apply**, and the red outline goes. ![Black Friday on the checkout map: the scenario panel on the left says two parts need more room, the email worker is outlined in red on the canvas, and the timeline along the bottom shades the hours it's short.](https://docs.flozu.app/_astro/scenario-light.D50hHthp_2r0uL7.webp) [Will it survive Black Friday?](https://docs.flozu.app/walkthroughs/black-friday.md) walks through the same thing with every number explained. > **Note** > > Flozu never runs or calls your systems. A scenario is worked out from the rates and limits on the map, so it’s only as good as those numbers. See [how traffic is worked out](https://docs.flozu.app/concepts/how-traffic-is-worked-out.md). ## Pick a scenario **Scenarios** lists the map’s own saved scenarios under **On this map**, then the ready-made ones, such as Normal day, Black Friday, Product launch, Back after an outage and Region down. See [ready-made scenarios](https://docs.flozu.app/reference/scenario-presets.md) for what each one does. **Scenario** at the top of the panel switches to another without going back to the list. Choosing **Scenario** in the mode menu at the foot of the canvas goes straight in. To end a scenario, choose another mode or close the panel. ## Read what happens The **What happens** tab opens on the outcome: how many parts need room and the busiest one’s share in big numbers, a bar with a segment for each part coloured by how it copes, and a sentence, such as “2 parts need more room on Black Friday”. Under it, **Fix it** lists the first three fixes. ![What happens for Black Friday: a card saying 2 of 3 parts need room with a coloured bar and the outcome, then Fix it with each part that needs more room and its fix, then a row each for Moments, Time and cost, Weak spots, The odds and What copes.](https://docs.flozu.app/_astro/needs-more-room-light.tXuLJjeI_2dF7W7.webp) Below that, each row has a headline, such as “4, trouble from 00:00” for **Moments**. Select a row to go in to it, and the way back at the top to come out. What you find in a row stays there while you look at something else. | Row | What it tells you | | ----------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Every fix** | Every fix, when there are more than three. | | **Moments** | The busiest and quietest moments, and when each part first gets tight, runs over, turns traffic away or fills its buffer. Select one to move the timeline there. | | **Needs more room** | Each part that’s ever short, worst first, at its own worst moment, with the fix sized for that moment, such as “Give it 126/s of capacity, 6.3 times today’s, or run 7 copies”. The **i** beside it says for how long and from when. | | **Time and cost** | Keep more copies running, let a part scale in, or make a part smaller when even its busiest moment uses under 30% of it. And [estimates](https://docs.flozu.app/guides/response-times-and-cost.md) of how long requests take and what the map costs. | | **Blast radius** and **Survive it** | For a scenario where something fails: what each failure touched, and what would carry on next time. See [failure drills](https://docs.flozu.app/guides/failure-drills.md). | | **Weak spots** | **Find weak spots** takes each part and zone down in turn at the busiest moment, and lists the ones that hurt most. | | **The odds** | For a scenario with random spikes: how often each part runs short over many runs. See [the odds](#the-odds). | | **What copes** | The parts with room to spare, and the ones with no capacity set. | Select a part in the panel to select it on the canvas and move the timeline to its worst moment. ## Fix what it finds Each fix has two buttons: - **Apply** makes the fix on the map as one change, so one undo takes it back. - **Try it** makes it in a [proposal](https://docs.flozu.app/guides/versions-and-proposals.md) instead: a copy of the map you can compare with the original, with the same scenario playing on both. The map itself doesn’t change until you adopt it. With more than one fix, **Try every fix in a proposal** makes them all in one, called “Cope with Black Friday”. A fix that changes the map’s shape, such as a standby copy, always starts as a proposal. ![A proposal called Cope with Black Friday on the checkout map, marked Copes now at the top: the scenario panel says all 3 checked parts cope on Black Friday, and the timeline sits at the day's busiest moment.](https://docs.flozu.app/_astro/black-friday-fix-light.U180IgwX_1G4hTv.webp) The scenario plays on the map as it stands, so changing a capacity in the inspector shows on the canvas straight away. ## On the canvas While a scenario runs, the canvas shows the moment at the timeline’s playhead: - Each line’s dots move at its rate then, and its label says the rate. - Each card’s figure counts the traffic then. - A part with a capacity or a limit has a **health strip** along the bottom of its card: how busy it is, as a share of what it can take, and a meter with a tick at the spare room to keep. Hover the strip to read it all in words. - Busy, critical and over parts are outlined in their colour, and an over part breathes. The lines into it turn red and slow down, as the work backs up. | Health | When | Colour | | ------------- | ----------------------------------------------------------------------------------------------------------------------- | ------ | | **Underused** | Under 30% busy, or under the spare room if that’s less. | Blue | | **Healthy** | Between underused and the spare room. | Green | | **Busy** | Past the spare room, 70% busy by default. | Yellow | | **Critical** | Past 90%, or past the spare room if that’s higher. | Orange | | **Over** | Past its capacity: **Overloaded**, **Queueing** when work backs up, or **Turning away** when its limit refuses traffic. | Red | | **Down** | A failure took it down, so nothing gets through. Its card fades behind a dashed outline. | Red | A part that fails some of its traffic, because something after it is down, adds the share that fails to its strip, such as “33% fail”. ## The timeline The timeline along the bottom draws how many people are online (or the multiple of today) across the scenario. An amber line along its top marks where a part is tight, a red one where it’s over, and diamonds mark the moments worth stopping at. The header says the time, the scenario, how busy it is then as a multiple of today’s busiest, and whether everything copes. ![The timeline for Black Friday: a curve of traffic across the day from 00:00 to 24:00, marked red along the top where parts need room, with the playhead at 00:10.](https://docs.flozu.app/_astro/timeline-light.B5yrEedO_ZNuc2f.webp) It opens at the scenario’s worst moment. To move through it: - **Drag** along it, or click anywhere on it. - Use the **arrow keys** for small steps, `Shift` with the arrows for bigger ones, and `Home` and `End` for the start and the end. `Space` plays and pauses. - **Skip** to the previous or next moment with the buttons on the right. - **Play** it through in about 30 seconds. The canvas follows the playhead. In a [failure drill](https://docs.flozu.app/guides/failure-drills.md), each failure has a bar along the foot of the graph, red for down and dashed amber for degraded, the stretch where traffic fails is shaded red, and the header says what’s wrong, such as “4 parts down, 33% failing”. ### Moments The moments marked on the timeline are the ones **Moments** lists, and a [story](https://docs.flozu.app/guides/stories.md) step can stop at any of them: | Moment | When | | ------------------------------- | ------------------------------------------------------------------------------------------------------- | | The busiest moment | Traffic at its highest. | | The quietest moment | Traffic at its lowest. | | _Part_ gets tight | That part first runs past its spare room. | | _Part_ runs over | That part first runs past its capacity. | | _Part_ turns traffic away | That part first turns traffic away at its limit. | | _Part_’s buffer fills | That queue or topic first fills up. | | _Part_ goes down | A failure takes it down, or a zone or several parts. A line _is cut_, and a degraded one _is degraded_. | | _Part_ moves traffic over | Traffic moves off what failed, onto a standby or the lines beside it. | | _Part_’s requests start failing | Some of what starts there first fails. | | _Part_ is back | The failure ends. | | Everything copes again | Everything is clear again. | ## Make your own The **Set up** tab shapes the scenario. Changing a saved or ready-made scenario turns it into a draft, shown as such in the picker. Say you’re running a TV ad at 8pm and expect three times the usual traffic for twenty minutes. Start from **Normal day**, and under **Events** add a **Spike** at 20:00 to a level of 3, holding for 20 minutes. Play it, and **What happens** says whether you’re ready. ### People online For each part where traffic starts, how many people use it at once on a normal day: at the busiest moment, at a typical one and at the quietest. Leave it out and a scenario counts in multiples of today. You can also set this in the part’s **Traffic** tab. ### Timeframe How long the scenario runs, from 10 minutes to 4 weeks; the time of day it starts; whether it follows a day’s rhythm (quietest at about 4am, busiest at about 8pm) or stays steady; and how busy it runs, as a multiple of today’s busiest moment. ### Events What happens to people over the timeframe. Each event has when it starts, its shape, and whether it reaches everyone or just one part where traffic starts, such as a push notification that only reaches the app. | Event | What it does | For example | | ---------- | ------------------------------------------------------------------ | ---------------------------------------------------- | | **Spike** | Rises to a level, holds, then falls back. | A TV ad, a newsletter, a ticket drop | | **Growth** | Grows to a level over a length of time, and stays there. | A launch that keeps growing for a week | | **Outage** | Nobody gets through for a while, then everyone comes back at once. | Your provider’s outage, then the rush when it’s back | Levels are multiples of a normal day’s busiest moment, so a map of 200 people and a map of 2 million both get a sensible Black Friday. Type lengths as “90 s”, “10 min” or “2 days”. ### Failures Take a part, a line or a whole zone down for a while, or leave it degraded with only part of what it can take. You can also right-click it on the canvas and choose **Take down here**. See [failure drills](https://docs.flozu.app/guides/failure-drills.md). ### Random spikes Up to how many random spikes to add, and the range of their peak, rise and hold. **Wobble** adds small moment-to-moment changes. A scenario always plays the same way until you choose **Roll again**, so you can show the same run to someone later. ### Advanced The **spare room** to keep, 30% by default (so a part aims to run at most 70% busy), and a plain steady multiple of today’s busiest for an hour. ## Save a scenario A draft belongs to this window only. **Save with the map** keeps it with the map, so it travels in files and share links, shows under the map in the **Maps** panel, and [stories](https://docs.flozu.app/guides/stories.md) can use it. Changes to a saved scenario save as you make them, each one step to undo. **Delete** removes it from the map and keeps showing it as a draft. ## The odds One run of a scenario with random spikes can be lucky. The odds play it many times, each with different spikes, and say how often each part runs short. 1. Open **The odds**, under **What happens**, for a scenario with random spikes or wobble. 2. Pick how many runs, from 20 to 200, and select **Try**. **Stop** ends them early. ![The odds section after 50 runs: everything copes in 0 of 50 runs, with Show the worst run, and the email worker short in 50 of 50 runs with a fix sized to cope in 95% of them.](https://docs.flozu.app/_astro/odds-light.ChUrOgMa_Z1p86SO.webp) The result says: - **How many runs everything copes in,** such as 41 of 50. - **How often each part runs short,** over its capacity or turning traffic away. - **The fix that copes in 95% of runs,** for each part that runs short, with its **Apply**. It’s sized from the run at the 95% mark when the runs are put in order of what that part needs. - **Show the worst run,** which sets the scenario to that run, so you can watch it on the timeline. A scenario with no random spikes and no wobble plays the same every time, so it has no odds to try. ## In presenting Presenting offers scenarios too, so a client can watch Black Friday play out. It leaves out the fixes, saving and people fields. See [presenting](https://docs.flozu.app/guides/presenting-and-sharing.md).