The first number is the unfiltered sensor stream: events per second times payload size.
Edge-Fog-Cloud Data Flow Animation
Trace IoT telemetry upward through edge, fog, and cloud tiers, then watch control commands return.
animation
architecture
edge
fog
cloud
data-flow
interactive-tool
Learner-ready edge-fog-cloud data flow workbench with scenario presets, animated upstream and downstream paths, raw-versus-processed mode, bandwidth reduction math, latency estimates, and mobile-safe learning support.
Interactive animation
Edge fog cloud
Data reduction math
Watch raw IoT telemetry become useful cloud data.
Devices create many small measurements. Edge gateways remove obvious noise, fog nodes combine local context, and the cloud keeps the history that is worth storing. Use the controls to compare a raw-to-cloud design with a processed path.
18.8 MB/s
Raw sensor stream
0.41 MB/s
Cloud ingress
97.8%
Traffic removed before cloud
11 ms
Control loop estimate
The edge tier drops duplicates, invalid readings, and values that do not need to leave the site.
The fog tier combines nearby devices, runs local rules, and sends smaller summaries upstream.
The cloud receives durable events, trends, model-training data, and fleet-level evidence.
Animated data path
Step through the upward telemetry path and the returning command path.
Device fleet
Sensors publish raw readings, events, and status changes. This is the highest-volume stream.
Events2400/s
Payload8 KB
Output18.8 MB/s
Edge gateway
Edge logic validates readings, removes duplicates, keeps immediate alarms local, and forwards the rest.
ActionFilter 82%
Local roleSafety loop
Output3.38 MB/s
Fog node
Fog services merge streams from a local site and convert many readings into fewer local insights.
ActionAggregate 88%
Local roleSite context
Output0.41 MB/s
Cloud platform
Cloud analytics store summaries, compare sites, train models, and return slower policy updates.
Ingress0.41 MB/s
Daily volume34.2 GB/day
Cloud roleHistory
Step 1: Devices publish raw telemetry
Raw rate = events per second x payload KB / 1024. The stream is still too expensive to send to the cloud unchanged.
1 of 5
Bandwidth evidence
Each bar is normalized to the raw stream. In raw mode, cloud ingress stays equal to raw traffic.
Raw rate
18.8 MB/s2400 events/s x 8 KB / 1024
Cloud volume
34.2 GB/dayAbout 1.00 TB/month at this ingest rate.
Reduction
97.8%Edge and fog together remove 18.34 MB/s before WAN upload.
Command loop
11 msDecision at edge: 3 ms up + 5 ms processing + 3 ms down.
Learning support and quick reference
Device fleet
Generates the raw stream. Devices usually have tight energy, memory, and radio constraints, so they should not do every possible analysis.
Edge tier
Runs closest to sensors or actuators. Use it for safety, privacy filtering, validation, compression, and fast local actions.
Fog tier
Sits at a local gateway, site server, or regional node. Use it when several devices need shared context without a cloud round trip.
Cloud tier
Stores durable history, compares sites, trains models, and coordinates fleet policy. It is powerful but usually slower and WAN-dependent.
Try this first
Press Step until the command returns. Watch the badge, highlighted tier, and formula cards change together.
Compare modes
Switch to Raw cloud path. The cloud bar should jump to the raw stream because the local reductions are bypassed.
Decision check
For a tight response target, move the command decision from edge to cloud and notice why cloud is not used for stop-now control loops.