Vibe Engines
YouTube
System Design

Design Google Maps

Step 1 / 9

Learn system design by building a mapping and navigation service like Google Maps step by step.

The numbers to beatz/x/ytile addressCDNedge cachedviewportonly fetched

The whole design, in writing

Learn system design by building a mapping and navigation service like Google Maps step by step. An interactive guide covering map tiles and CDN delivery, the road graph, shortest-path routing with A* and contraction hierarchies, traffic-aware ETAs, geocoding, and scaling.

Every step of the build above, written out: the problem each piece solves, the option that was taken and the ones that were not, the numbers, and how it fails in production.

The big idea

What is Google Maps?

Render the whole world as a smooth, zoomable map, and compute the fastest route between any two points — accounting for live traffic — in well under a second, for billions of users. Two very different problems live here: drawing the map and routing over it.

Userpan · route
New in this step: User.

Pre-render the map into cacheable tiles served from the edge, and model roads as a graph you run shortest-path search over. Make routing fast with precomputation, fold in live traffic for ETAs, and add geocoding to connect names to places.

What the new pieces do

Userclient
Pans and zooms the map, searches places, and asks for directions with a live ETA — all expected to feel instant while moving.

Step 1 · Draw the world

Map tiles

You can’t send the entire planet’s map to a phone, nor re-render it per request. Yet panning and zooming must feel instant at every zoom level, everywhere.

Maps APITile ServiceTile Store + CDN
New in this step: Maps API, Tile Service, Tile Store + CDN. · swipe to pan the diagram

You can’t send the planet’s map to a phone or re-render it per request. How do you make pan/zoom feel instant everywhere?

  1. Rendering on demand for billions of pans/zooms is enormously expensive and barely cacheable since every viewport differs. Precompute reusable pieces instead.

  2. Small tiles addressed by zoom/x/y let the client fetch only what’s in view; tiles rarely change, so they edge-cache beautifully and load fast worldwide. "Render the world" becomes "fetch a few static images."

  3. The planet’s map data is far too large to download to a phone, and would be stale immediately. The client should request just the tiles in its viewport, rendered ahead of time.

Pre-render the map into tiles — small square images (or vector tiles) addressed by zoom / x / y. The client requests only the tiles in view; a Tile Service serves them from a CDN-fronted store. Since tiles rarely change, they cache beautifully.

  • z/x/ytile address
  • CDNedge cached
  • viewportonly fetched

What the new pieces do

Maps APIbackend
Routes the three core requests: map tiles to render, route computations, and place/address lookups. The front door to every map feature.
Tile Servicecache
Serves the little square images (or vector tiles) that tile together into the map, addressed by zoom level and x/y grid coordinates.
Tile Store + CDNstore
Pre-rendered tiles in cheap storage, served from a CDN. Tiles change rarely, so they cache at the edge and load fast worldwide.

Step 2 · Roads as math

The road graph

Directions aren’t about images — they’re about connectivity. To compute a route you need a structure that captures which roads connect to which, and how “costly” each segment is.

Maps APIRoad Graph
New in this step: Road Graph. · swipe to pan the diagram

Directions aren’t about images — they’re about connectivity. What structure do you compute routes over?

  1. Tiles are pixels for looking at, not connectivity for computing over — you can’t route on images. Drawing and routing are different problems needing different structures.

  2. A flat list of roads doesn’t capture which connects to which, turn restrictions, or costs — everything routing needs. You need explicit connectivity with weights.

  3. Nodes are intersections, edges are road segments weighted by length and typical speed (plus one-ways and turn restrictions). Routing becomes shortest-path over this graph — separate from the visual tiles.

Model the map as a graph: intersections are nodes, road segments are edges weighted by length and typical speed (plus one-ways and turn restrictions). Routing becomes a shortest-path problem over this graph.

What the new pieces do

Road Graphstore
The map as a graph: intersections are nodes, road segments are weighted edges (by length/speed). All routing is graph search over this.

Step 3 · Find the way

Shortest-path routing

Given start and end nodes, you need the lowest-cost path through a graph with hundreds of millions of edges — and a plain Dijkstra exploration would crawl outward over an enormous area.

Maps APIRouting EngineRoad Graph
New in this step: Routing Engine. · swipe to pan the diagram

Find the lowest-cost path in a graph with hundreds of millions of edges. Plain Dijkstra crawls outward everywhere. Better?

  1. Enumerating all paths is exponential — infeasible on a continental graph. You need a guided search that provably finds the optimum without exploring everything.

  2. The heuristic (as-the-crow-flies distance to the goal) steers the search toward the destination instead of expanding in all directions — far fewer nodes explored, same optimal path.

  3. Pure greedy best-first isn’t optimal — it can charge toward the goal down a dead-end or slow road. A* combines goal-direction WITH actual cost-so-far to stay optimal.

A Routing Engine runs A*: Dijkstra guided by a heuristic (straight-line distance to the goal) so the search heads toward the destination instead of expanding in all directions. Far fewer nodes explored, same optimal path.

What the new pieces do

Routing Engineservice
Computes the best route between two points over the road graph, accounting for road types, turn restrictions and, later, live traffic.

Back of the envelope

A* = Dijkstra + heuristic
straight-line distance to the goal
heuristic steers toward destination
a fraction of the nodes explored
admissible ⇒ still optimal
never overestimates real distance

Step 4 · Make it instant

Precompute for speed

Even A* is too slow for continental routes at query time, and you serve millions of route requests per second. Exploring the raw graph on every request simply won’t hit sub-second latency.

UserMaps APITile ServiceTile Store + CDNRouting EngineRoad Graph
The system as it stands at this step. · swipe to pan the diagram

Even A* is too slow for continental routes at millions of queries/sec. How do you get millisecond long-distance routes?

  1. More machines lower cost-per-query but each continental A* is still slow — you can’t parallelize one path search enough for sub-second. Do less work per query, not more hardware.

  2. There are astronomically many origin-destination pairs — precomputing all routes is impossible to store. You precompute reusable shortcuts, not full answers.

  3. Heavy offline preprocessing builds "highway" shortcut edges; online queries hop over them, skipping most local roads. Long routes become a few hops at the top of a hierarchy — milliseconds. The offline/online split again.

Precompute shortcuts with contraction hierarchies (or highway/transit hierarchies): collapse the graph so long routes hop over precomputed “highways” instead of every local road. Queries then run in milliseconds by skipping most of the graph.

  • offlineshortcuts built
  • msroute query
  • continentalin real time

Back of the envelope

offline: build shortcut "highways"
heavy preprocessing, done once
online: a few hops at the top
skip most local roads
continental route ⇒ ms
trade build time for query time

Step 5 · Account for traffic

Live ETAs

The shortest route by distance isn’t the fastest in a jam, and a static ETA is wrong the moment congestion changes. You need to know current speeds on every road.

ETA Servicetraffic-awareTraffic Datalive speedsLocation PingsGPS stream
New in this step: ETA Service, Traffic Data, Location Pings.

The shortest route by distance isn’t fastest in a jam, and you need CURRENT speeds on every road. Where do they come from?

  1. Physical sensors on every road on earth are impossibly expensive and sparse. You already have hundreds of millions of moving sensors — the users’ phones.

  2. Every moving user is a sensor; aggregating anonymized location pings gives live speed per segment. The ETA service weights routes by these (plus historical patterns), and routing re-weights edges to avoid jams.

  3. Time-of-day priors help (and are a fallback), but can’t see today’s accident or event-driven jam. Real ETAs need live ground truth, which the crowd of moving users provides.

Phones stream anonymized Location Pings; aggregating them gives live Traffic Data — current speed per segment. The ETA Service weights the route by these speeds (and historical patterns) to estimate arrival, and routing re-weights edges to avoid jams.

What the new pieces do

ETA Serviceservice
Turns a route into a time estimate using live and historical speeds per segment, so the ETA reflects actual conditions, not just distance.
Traffic Datastore
Current speed per road segment, aggregated from millions of anonymized GPS pings. Feeds both ETAs and re-routing around jams.
Location Pingsbus
A firehose of anonymized location updates from phones in motion. Aggregated into live traffic speeds that keep ETAs and routes honest.

Back of the envelope

every moving phone = a sensor
100Ms of anonymized pings
aggregate ⇒ live speed / segment
real-time ground truth
re-weight edges ⇒ avoid jams
fastest, not just shortest

Step 6 · Names to places

Geocoding

Users type “Eiffel Tower” or “221B Baker St”, not latitude/longitude. The router and map need coordinates, and search results need human-readable addresses back.

Maps APItiles + routesGeocodingaddress ↔ latlng
New in this step: Geocoding.

Users type "Eiffel Tower" or "221B Baker St", but the router and tiles need coordinates. How do you bridge them?

  1. Forcing coordinates is hostile to users who think in names and addresses. The system must translate human input into coordinates, not push that burden onto people.

  2. Graph nodes are intersections, not a searchable place/address index — and addresses don’t map cleanly to nodes. Place lookup is its own retrieval problem with ranking and autocomplete.

  3. Geocoding (and reverse geocoding) translates names/addresses to lat-lng and back, backed by a places index with autocomplete and relevance/distance ranking — feeding clean coordinates into routing and the map.

A Geocoding service translates addresses and place names into coordinates (and reverse-geocodes coordinates into addresses), backed by a places index. It’s the bridge between what users type and what the graph and tiles understand.

What the new pieces do

Geocodingservice
Translates a typed address or place name into coordinates (and back), so “coffee near me” becomes points the map and router can use.

Step 7 · Scale the planet

The sharp edges

Billions of tile and route requests, a graph too big for one machine, and traffic that shifts by the minute. No single server holds the world or serves that load.

UserMaps APITile ServiceTile Store + CDNRouting EngineRoad GraphETA ServiceTraffic DataGeocodingLocation Pings
The system as it stands at this step. · swipe to pan the diagram

Lean on the CDN for tiles (most requests never hit your origin), partition the road graph by region with cross-region links for long routes, and continuously update edge weights from the traffic stream. Cache popular routes and precomputed hierarchies per region.

You did it

You just designed Google Maps.

UserMaps APITile ServiceTile Store + CDNRouting EngineRoad GraphETA ServiceTraffic DataGeocodingLocation Pings
The finished design, end to end. · swipe to pan the diagram

Everything you assembled, in order

  • Pre-rendered z/x/y map tiles served from a CDN make browsing instant.
  • Roads modeled as a weighted graph of intersections and segments.
  • A* routing (Dijkstra + heuristic) finds optimal paths efficiently.
  • Contraction hierarchies precompute shortcuts for millisecond long-distance routes.
  • Crowdsourced GPS pings aggregate into live traffic for honest ETAs and re-routing.
  • Geocoding bridges typed addresses/places to coordinates and back.
  • CDN tiles plus region-partitioned graphs scale to billions of requests.

Where an interviewer pokes next

Getting the boxes right is the easy half. These are the questions that separate a candidate who drew the diagram from one who has run the thing. Answer each one out loud before you open it.

  1. Why contraction hierarchies over just A*?

    A* prunes the search but still explores a swath proportional to route length, so a cross-continent query touches millions of nodes — too slow at scale. Contraction hierarchies do heavy offline preprocessing to add shortcut edges that let a query "hop" over entire regions, reducing a long route to a handful of edges answered in milliseconds. The classic trade: expensive build, cheap query, justified because the graph changes slowly and queries are constant.

  2. How does live traffic get folded in without rebuilding everything?

    Edge weights update continuously from the traffic stream, but contraction-hierarchy shortcuts are expensive to rebuild — so production uses customizable route planning (CRP) or live overlays: keep the static hierarchy, apply a fast "metric customization" layer that re-weights affected edges/shortcuts when traffic changes. Routing reads current weights; you re-customize periodically rather than re-contracting the whole graph.

  3. How are ETAs made accurate, not just current-speed × distance?

    Multiple signals: live per-segment speeds for the near term, historical speed-by-time-of-day for parts you’ll reach later, plus learned corrections (turn delays, light patterns, congestion build-up). The ETA predicts conditions along the trip’s timeline, not a single snapshot — which is why a 9am estimate differs from 9pm on the same route.

  4. How do you keep the map and graph fresh as roads change?

    Two pipelines: tiles are re-rendered offline and pushed to the CDN as new versions (old ones expire), and the road graph is updated from map data (new roads, closures, turn restrictions) then re-preprocessed for hierarchies region by region. Both are eventually-consistent batch jobs; urgent changes (a closure) apply as live edge-weight overrides ahead of the full rebuild.

  5. How would you scale to billions of tile + route requests?

    Geography is a natural shard key. Tiles are almost entirely served from the CDN edge (most requests never reach origin), and the road graph is partitioned by region with cross-region links for long routes, so a query mostly hits one region. Cache popular routes and per-region hierarchies. Most routes are local, so partition-by-space plus aggressive edge caching carries the planet-scale load.

Check yourself — the answers, and why

Eight steps in, these are the calls you should be able to make cold. Pick one, then read why.

  1. Map browsing is fast because the world is…

    • Rendered per request
    • Pre-rendered into z/x/y tiles served from a CDN
    • Sent fully to the client

    Immutable, edge-cached tiles turn "render the world" into "fetch a few static images near the viewport".

  2. Routing runs over…

    • The map tiles
    • A weighted graph of intersections (nodes) and segments (edges)
    • A list of addresses

    Drawing and routing are separate: tiles for looking, a weighted graph for computing paths.

  3. A* beats plain Dijkstra by…

    • Exploring more nodes
    • Using a heuristic to steer the search toward the goal
    • Caching every route

    An admissible straight-line heuristic prunes the search toward the destination while staying optimal.

  4. Continental routes return in milliseconds thanks to…

    • Faster servers
    • Contraction hierarchies — offline shortcuts queries hop over
    • Smaller maps

    Trade heavy offline preprocessing for cheap online queries that skip most of the graph.

  5. Live traffic data comes from…

    • Road sensors everywhere
    • Aggregated anonymized GPS pings from moving phones
    • Time of day alone

    Every moving user is a sensor; aggregated motion becomes per-segment live speeds for honest ETAs.

How you’d open this design in an interview

Before any boxes: agree what it must do, pin the qualities that shape everything, then build — naming each trade-off as you make it. The walkthrough above is that exact order.

What it must do

Agree on these before drawing a single box.

  • Draw the map: pan and zoom a smooth world map at every zoom level.
  • Route: compute the fastest path between two points over the road network.
  • ETA: a traffic-aware time estimate that reflects current conditions.
  • Search places: turn "Eiffel Tower" or an address into coordinates and back.
  • Re-route: steer around jams as live traffic changes.

The qualities that shape everything

Each one names the mechanism that buys it.

Instant pan/zoom worldwide
Pre-render the map into immutable z/x/y tiles served from a CDN, so the client fetches only the viewport and most requests never reach the origin.
Routing you can actually compute
Model roads as a weighted graph — intersections are nodes, segments are edges weighted by length and speed — separate from the visual tiles, turning directions into shortest-path search.
Optimal path without crawling the whole graph
A* — Dijkstra guided by a straight-line-distance heuristic — steers toward the destination, exploring a fraction of the nodes while staying optimal.
Millisecond continental routes
Contraction hierarchies precompute shortcut "highways" offline so a long query hops over most local roads — trading heavy build time for cheap queries.
ETAs that reflect real conditions
Aggregate anonymized GPS pings into live per-segment speeds; the ETA service weights the route by them and routing re-weights edges to avoid jams.
Bridge what users type to what the graph understands
A geocoding service translates addresses/place names to coordinates and back, via a places index with autocomplete and relevance ranking.

The trade-offs you say out loud

Senior signal isn’t the boxes — it’s naming what you gave up and why it was the right price.

Pre-rendered z/x/y tiles over rendering each viewport per request

Rendering on demand for billions of pans is enormously expensive and barely cacheable since every viewport differs. Immutable tiles turn "render the world" into "fetch a few static images," and they edge-cache beautifully.

A weighted road graph over routing on the display tiles

Tiles are pixels for looking at, not connectivity for computing over — you can’t route on images. A graph of intersections and weighted segments encodes distance, speed and legal turns for shortest-path search.

A* with a distance heuristic over plain Dijkstra outward search

Dijkstra expands in all directions over a continental graph. An admissible straight-line heuristic steers the search toward the goal, touching a fraction of the nodes while still returning the optimal route.

Contraction hierarchies (offline) over caching every origin-destination route

There are astronomically many O-D pairs — precomputing all routes is impossible to store. Precompute reusable shortcut edges instead, so long routes become a few hops at the top of the hierarchy in milliseconds.

Crowdsourced GPS pings over sensors on every road

Physical sensors on every road on earth are impossibly expensive and sparse. Hundreds of millions of moving phones are already sensors — aggregate their anonymized motion into live per-segment speeds.

What this teaches

Learn system design by building a mapping and navigation service like Google Maps step by step. An interactive guide covering map tiles and CDN delivery, the road graph, shortest-path routing with A* and contraction hierarchies, traffic-aware ETAs, geocoding, and scaling.

Key takeaways

  • Pre-rendered z/x/y map tiles served from a CDN make browsing instant.
  • Roads modeled as a weighted graph of intersections and segments.
  • A* routing (Dijkstra + heuristic) finds optimal paths efficiently.
  • Contraction hierarchies precompute shortcuts for millisecond long-distance routes.
  • Crowdsourced GPS pings aggregate into live traffic for honest ETAs and re-routing.
  • Geocoding bridges typed addresses/places to coordinates and back.
  • CDN tiles plus region-partitioned graphs scale to billions of requests.

Concepts covered

  • What is Google Maps?
  • Map tiles
  • The road graph
  • Shortest-path routing
  • Precompute for speed
  • Live ETAs
  • Geocoding
  • The sharp edges
built to be navigated, not memorized — make the calls, kill the traffic feed, run the gauntlet.
Finished this one? 0 / 65 System Designs done

Explore the topic

See this alongside everything else on the same subject — handbooks, system designs, challenges and tools, in one place.

More System Designs