Your system is set to reduce motion, so the data centre is paused. This screen has a slowly orbiting camera, moving packets, and a pulsing red glow at the screen edges during incidents.
This page is the operations wall of an invented cloud company, built around a data centre drawn as a small city. Sixty-four server towers stand on four glowing platforms, one per region, around a tall central core. Each tower rises and falls with its load, its windows light up as it gets busier, and packets of light run along the floor between the towers and the core. The camera turns slowly around the city. About every four minutes something goes wrong: an alarm fires, the camera dives onto the tower under attack, red streaks rain down on it, and a green shield dome rises over its region until the incident is resolved. Around the scene sit the panels of a real-looking monitoring dashboard. Click ⛶ FULLSCREEN and it fills the screen.
No. It is an animation, and the bar under it says SIMULATION, as does a tag in the top corner of the screen. The company name, the four regions, the 64 towers, the eight services, every number on the panels, the incident IDs, and every line of the event stream are invented and calculated in your browser. The page makes no connections while it runs and monitors nothing. The clock in the header is a story clock that starts at 14:11:16, not your local time.
Real data centres do not look like this from the inside. They are long, loud, chilly halls of identical metal cabinets. The section on how real data centres work below explains what the towers, the core, and the shield stand for.
The ten-hour cycle holds 150 incidents, one in each four-minute slot, so the gap between two incidents varies from about two and a quarter minutes to about six. Most incidents last between 70 and 115 seconds, and each one plays out in four steps shown on the threat response tracker.
There are six incident types, each with its own mitigation and affected service: a DDoS attack (scrubbing traffic at the edge; edge-cdn), an intrusion attempt (isolating the host segment; auth-service), a ransomware beacon (blocking command-and-control domains; object-storage), a brute-force campaign (rate limiting the auth API; auth-service), a BGP route leak (withdrawing bad prefixes; api-gateway), and a database failover (promoting a replica; db-primary). The shield dome comes up for every type, which is the screen's visual shorthand for "defences engaged" rather than a picture of any real mechanism.
Now and then a service also drops to amber for a short spell of 8 to 38 seconds, which nudges the threat level to GUARDED when no incident is running.
Moments worth seeing in the first part of the cycle (seek with the bar or use ?t=):
| Time | Seconds | What happens |
|---|---|---|
| 00:00:00 | 0 | Start-up: the panels wipe on and the towers rise from the floor, from the core outwards, in about three seconds |
| 00:00:12 | 12 | First incident, a DDoS attack on EUW-07 in EU-WEST; the camera dives in |
| 00:00:27 | 27 | The shield dome rises over EU-WEST |
| 00:00:42 | 42 | Resolved; the camera is back in orbit by about 1:02 |
| 00:06:15 | 375 | BGP route leak on NAE-03 in NA-EAST |
| 00:09:52 | 592 | Brute-force campaign on NAE-13; the dome rises at about 10:30 |
| 00:12:34 | 754 | Second DDoS attack, on EUW-11 |
| 00:25:51 | 1551 | Database failover on LAS-02 in LATAM |
| 00:33:08 | 1988 | First intrusion attempt, on EUW-07 |
| 01:01:56 | 3716 | First ransomware beacon, on APS-08 in AP-SOUTH |
Over the full ten hours there are 150 incidents: 35 BGP route leaks, 29 DDoS attacks, 28 database failovers, 24 brute-force campaigns, 21 intrusion attempts, and 13 ransomware beacons. AP-SOUTH is hit 43 times, LATAM 40, EU-WEST 37, and NA-EAST 30. An incident is in progress about 39% of the time. There are also 327 short service slowdowns, placed at random.
?t= and the number of seconds to the address, for example /data-center?t=592 for the first brute-force campaign. Minutes and hours also work: ?t=25m51s or ?t=1h30m.The page draws at the full resolution of your screen, up to 4K. On a slow computer it lowers the resolution by itself to keep the motion smooth.
Background screens for films, series, and adverts set in a tech company, a control room, or a server hall; a wall display for an office, a reception area, a trade-show stand, or a hackathon; a backdrop for videos and talks about cloud computing, networking, or cyber security; an ambient screen for coding and tech streams; and a "the servers are under attack" moment in escape rooms and live games. It moves calmly most of the time and gets busy every few minutes, so it holds attention without tiring the eye. For OBS setup as a browser source, see the streaming toolkit.
A data centre is a building whose job is to keep computers running: to give them power, keep them cool, connect them to the network, and keep them safe. Websites, apps, email, and cloud storage all run on servers in places like this. The screen turns each group of servers into a tower and the network into lines of light, but the ideas behind it are real.
Servers are flat metal boxes bolted into standard 19-inch racks, tall cabinets measured in rack units (one unit, 1U, is 1.75 inches or 44.45 mm); a full-height rack is commonly 42U. Racks stand in long rows on a raised floor or a concrete slab, with network and power cables running overhead or underneath. A large facility has several halls, each with hundreds of racks. The towers on this screen are closer to a row or a pod of racks than to a single server, and the floor traces stand for the network fabric that links them.
Power is the first thing a data centre is designed around. Electricity usually arrives from the grid on more than one feed. Inside, an uninterruptible power supply (UPS), a large bank of batteries, carries the load the instant the grid fails, for the seconds to minutes it takes the diesel generators to start and take over. Servers often have two power supplies, each plugged into a separate "A" and "B" power path, so a fault on one path does not switch anything off.
Almost every watt a server uses turns into heat, so cooling is the other half of the design. The usual layout is hot aisle / cold aisle: racks face each other in pairs so that servers draw cool air in from a shared cold aisle at the front and blow hot air out into a hot aisle at the back. Containment panels keep the two apart, and air handlers cool the hot air and send it round again. Many sites use outside air or water for "free cooling" when the weather allows. Racks full of AI accelerators produce so much heat that they are increasingly cooled with liquid piped to the chips; the AI data center screen shows a hall built for that kind of work.
Power usage effectiveness (PUE) is the standard measure of a data centre's overhead. It is the total energy used by the facility divided by the energy used by the IT equipment. A PUE of 1.0 would mean every watt reaches the computers; a PUE of 2.0 means as much power again goes on cooling, power conversion, and lighting. Industry surveys have put the average at roughly 1.5 to 1.6 for several years, while the most efficient large sites report figures close to 1.1. The measure says nothing about how useful the computing is, only how much is spent keeping it running.
Data centres are built so that any single part can fail. The language for this counts spare capacity against N, the amount needed to carry the load: N+1 means one spare unit (one more generator or cooling unit than needed), and 2N means two complete, independent systems. The Uptime Institute's widely used tier scale grades whole facilities:
| Tier | Name | What it means |
|---|---|---|
| I | Basic capacity | Dedicated power and cooling, but one path and no spares; maintenance means a shutdown |
| II | Redundant capacity components | Spare generators, UPS modules, and cooling units, still on a single distribution path |
| III | Concurrently maintainable | Any component or path can be taken out for maintenance without shutting down the IT equipment |
| IV | Fault tolerant | Can ride through any single unplanned failure without affecting the IT equipment |
No single building is safe from fire, flood, or a cut fibre, so large services spread out. Cloud providers group data centres into regions (the screen's NA-EAST and EU-WEST are invented names in that style), and each region into several availability zones, separate facilities with their own power, cooling, and network, far enough apart that one local disaster should not take out two of them. Data is copied between them, which is what the pink replication packets stand for. When one part fails, traffic moves to another: a load balancer stops sending requests to a server that fails its health checks, and a database replica is promoted to take over from a primary that has stopped responding. The "database failover" incidents on the screen show that last case.
None of this is visible as streaks of light. Real teams see it as spikes on graphs, alerts, and log lines, much like the panels around the scene.
The four steps on the threat response panel follow the way real operations teams run an incident. Monitoring raises an alert and pages the engineer on call. Someone takes charge, the team works out what is happening, applies a fix or a workaround, and watches to make sure it holds before declaring the incident resolved. Many companies publish these steps on a public status page with words such as investigating, identified, monitoring, and resolved. Afterwards the team writes a review of what happened and what to change, ideally without blaming individuals. The difference is time: on this screen an incident takes a minute or two, while a real one can take an afternoon.
The whole ten-hour cycle is worked out in advance from a fixed random seed: when each incident starts, how long it lasts, which tower it hits, and when each service has a slow spell. The metrics are sums of slow sine waves with a little seeded noise, so they wander believably but always come out the same at the same moment. The city is drawn with a parallel projection, the flat "isometric" look of strategy games where distant things are not drawn smaller, and the towers are painted from back to front so that near ones cover far ones. One packet is launched every 0.11 seconds, and its route, colour, and arrival are all derived from its number. The camera turns a quarter circle every 90 seconds, drifting slowly and then swinging round, and during an incident it eases in towards the target. To protect screens left on for hours, the whole picture also wanders by a few pixels over thirteen and nineteen minutes. Because every frame is a function of the time alone, the page can jump anywhere in the cycle at once.
No. It is an animation. The company, the 64 towers, the regions, the services, the metrics, the incidents, and every line of the event stream are invented and drawn in your browser. The page connects to nothing and monitors nothing.
NA-EAST, EU-WEST, AP-SOUTH, and LATAM are invented names in the style that cloud providers use for their regions: a continent and a direction. Each region on the screen is a platform of sixteen towers, and each tower stands for a group of servers. Real cloud regions are separate groups of buildings, usually in different cities or countries.
Ten hours. The cycle holds 150 incidents, fifteen an hour, each on its own tower with its own timing. After ten hours it starts again from the beginning. The screen is calculated frame by frame rather than played from a recording, so it needs no download and never buffers.
Yes. Drag the seek bar, or press the left and right arrow keys to move ten seconds at a time. To open the page at a given moment, add ?t= to the address with the number of seconds, for example ?t=12 for the first incident or ?t=592 for the first brute-force campaign.
A network operations centre (NOC) watches whether systems are up and fast: outages, slow services, full disks, failed links. A security operations centre (SOC) watches for attacks: intrusions, malware, stolen accounts. This screen mixes the two, as film screens usually do. In a real company they are often separate teams that work closely during an incident.
Yes. It is free to use for videos, films, streams, presentations, and events. Click FULLSCREEN, or add the page as an OBS browser source at 1920 × 1080 or 3840 × 2160; the scene is drawn at the full resolution of the source.
There are no full-screen flashes or fades. During an incident a red glow builds around the edges of the screen and swells gently on a three-second cycle until the shield goes up. The roof of the attacked tower pulses about once a second, and small sparks mark where hostile streaks land. At the very start of the cycle the panels wipe on over about three seconds. If your system is set to reduce motion, the page opens paused. See accessibility and visual effects for more.
Once the page and its fonts have loaded, the simulation makes no further network requests, so it keeps running if the connection drops. The optional VIDEO view streams from YouTube and needs a connection.
The AI data center screen shows a hall of GPU racks training a model, and the SOC screen follows a security team through a long attack campaign. The futuristic tech dashboard is a flatter control panel in the same spirit, and the cyber attack map shows attacks flying between countries. To learn what the DDoS incidents stand for, read what a DDoS attack is.