Vibe Engines
YouTube
System Design

Design a Web Crawler

Step 1 / 9

Learn system design by building a distributed web crawler step by step.

The numbers to beatbillionsURLs to trackO(1)seen check~bitsper URL

The whole design, in writing

Learn system design by building a distributed web crawler step by step. An interactive guide covering the URL frontier, the fetch–parse–enqueue loop, dedup with bloom filters, politeness and robots.txt, DNS caching, content storage and dedup, and sharding the frontier.

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 a web crawler?

A crawler downloads the web — start from a few URLs, fetch each page, find its links, fetch those too, forever. Trivial as a weekend script; brutal at billions of pages, where you must be fast, polite, dedup-aware, and never get stuck in a loop or a trap.

Seed URLsstarting set
New in this step: Seed URLs.

Model it as one big loop over a queue: pull a URL, fetch it, extract links, enqueue the new ones. Then make that loop scale, behave, and stay efficient. Everything we add defends that simple loop against the messy reality of the web.

What the new pieces do

Seed URLsinput
The initial list of URLs to crawl from — a few well-connected pages from which the rest of the reachable web unfolds, link by link.

Step 1 · The loop

Fetch, parse, enqueue

At its heart a crawl is a graph traversal: you have URLs to visit, you visit one, it reveals more URLs, repeat. You need somewhere to hold the “to-visit” list and workers to do the visiting.

URL Frontierwhat to crawl nextFetcher Workersdownload pagesParserextract links
New in this step: URL Frontier, Fetcher Workers, Parser.

A crawl reveals more URLs as it fetches. How do you structure the loop?

  1. Deep recursion blows the stack, can’t be distributed, and couples fetch to parse so neither scales. The to-visit set needs to be explicit state, not the call stack.

  2. A single sequential worker can’t approach billions of pages, and mixes network-bound fetching with CPU-bound parsing. You need a shared queue and many workers.

  3. The frontier holds what to crawl next; fetchers download, the parser extracts links and pushes new ones back. Decoupling fetch (network-bound) from parse (CPU-bound) lets each scale on its own.

A URL Frontier holds what to crawl next. Fetcher workers pull from it and download pages; the Parser extracts links and pushes the new ones back into the frontier. Seeds prime the loop and it runs itself from there.

What the new pieces do

URL Frontierbackend
The brain of the crawler: the queue of URLs waiting to be fetched. Its ordering decides what gets crawled, how soon, and how often.
Fetcher Workersworker
Pull URLs from the frontier and download the HTML over HTTP. A large, mostly-waiting-on-the-network fleet — the crawler’s muscle.
Parserworker
Reads each downloaded page, pulls out the links, normalizes them, and feeds the new ones back into the frontier — that loop is the crawl.

Step 2 · Don’t crawl forever

Have I seen this URL?

The web is a cyclic graph — pages link back and forth endlessly. Without memory, the crawler re-queues the same URLs over and over and never makes progress, drowning in duplicates.

URL FrontierFetcher WorkersParserSeen Set
New in this step: Seen Set. · swipe to pan the diagram

The web is cyclic — pages link back and forth. How do you avoid re-crawling forever, at billions of URLs?

  1. An exact set of billions of URLs is hundreds of GB — too big for memory on every frontier node. You need a far more compact "have I seen this?" structure.

  2. A bloom filter answers "probably seen / definitely not" in O(1) using a few bits per URL. It may rarely skip a new URL but never re-crawls a seen one — "mostly right and tiny" beats "exact and enormous".

  3. Without an up-front seen-check the crawler loops on cycles and never makes forward progress — it drowns in duplicates before any dedup stage runs. Termination has to happen at enqueue time.

Before enqueuing a link, check a Seen Set. At billions of URLs, an exact set is too big for memory, so use a bloom filter: O(1) “probably seen / definitely not seen” in a fraction of the space, backed by a durable set for certainty.

  • billionsURLs to track
  • O(1)seen check
  • ~bitsper URL

What the new pieces do

Seen Setstore
Remembers which URLs have already been queued so the crawler doesn’t loop forever. A bloom filter answers “seen this?” in O(1) using tiny memory.

Back of the envelope

billions of URLs, exact set ⇒ 100s of GB
too big to hold in memory per node
bloom filter ⇒ ~a few bits / URL
O(1) check in a fraction of the space
rare false positive
may skip a new URL, never re-crawls a seen one

Step 3 · Be a good citizen

Politeness & robots.txt

Thousands of fetchers all hammering one small site at once is indistinguishable from an attack — you’ll knock it over and get your crawler banned. And many sites declare what you may and may not crawl.

URL Frontierwhat to crawl nextFetcher Workersdownload pagesParserextract linksPolitenessrobots · rate/host
New in this step: Politeness.

Thousands of fetchers could hammer one small site at once. How do you not look like an attack?

  1. A global cap still lets all that traffic pile onto one unlucky host while big sites are under-crawled. Politeness is a per-domain promise, not a global throttle.

  2. Maximum speed per host is exactly the DDoS that knocks small servers over and gets your IPs banned — ending your crawl. Speed must be bounded per host.

  3. Obey each site’s rules and pace requests per domain, while crawling thousands of hosts simultaneously. Gentle on each, fast in aggregate — and the reason hosts become the partition unit later.

Add a Politeness layer: honor each host’s robots.txt, and enforce a per-host rate limit with a delay between requests to the same domain. Group the frontier by host so one site’s crawl is paced independently of others.

  • robots.txtobeyed
  • 1/hostrate limited
  • parallelacross hosts

What the new pieces do

Politenessguard
Obeys each site’s robots.txt and caps requests per host, so the crawler is a good citizen and doesn’t accidentally DDoS a small server.

Step 4 · The hidden bottleneck

DNS at scale

Every fetch needs a hostname resolved to an IP. DNS lookups are slow (tens of ms) and, at billions of requests, become the dominant cost — and can hammer DNS servers as hard as you hammer websites.

Fetcher Workersdownload pagesDNS Cachehost → IP
New in this step: DNS Cache.

Every fetch resolves a hostname to an IP, and DNS is slow (tens of ms). At billions of fetches, what breaks?

  1. At billions of requests, tens-of-ms lookups become the dominant cost and can hammer DNS servers as hard as you hammer sites. Ignoring DNS makes it the silent bottleneck.

  2. Most pages on a site share one IP, so the first lookup serves thousands of fetches; async resolution keeps fetchers from blocking. The unglamorous infra often decides throughput more than scheduling.

  3. IPs change and rotate (CDNs, load balancers), so hard-coding breaks constantly and silently fetches the wrong server. You need caching with a TTL, not static mappings.

Put a DNS Cache in front of resolution and resolve asynchronously so fetchers don’t block. Cache by host with a TTL; most pages on a site share one IP, so the first lookup serves thousands of fetches.

What the new pieces do

DNS Cacheservice
Resolving hostnames is slow and would dominate latency at billions of fetches. A cache (and async resolver) keeps lookups from becoming the bottleneck.

Back of the envelope

lookup ≈ tens of ms
dominant per-fetch cost at billions of fetches
1 host ⇒ ~1 IP
first lookup serves thousands of fetches
cache + async resolve
keeps the fetch pipeline full, fetchers unblocked

Step 5 · Keep what you fetched

Store & dedup content

Downloaded pages must be stored for downstream indexing — but the web is full of duplicate content: mirrors, print versions, session-ID URLs. Different URLs, identical pages. Storing them all wastes enormous space and indexing effort.

Fetcher Workersdownload pagesParserextract linksSeen Setbloom filterDNS Cachehost → IPContent Storeraw HTMLContent Hashesnear-dup detect
New in this step: Content Store, Content Hashes.

Mirrors, print versions and session-ID URLs deliver identical pages under different URLs. How do you not store the web twice?

  1. The bloom filter dedups URLs, not content — a brand-new URL can still return a page you’ve stored a thousand times. URL-dedup and content-dedup are different jobs.

  2. Comparing against the entire corpus per page is impossibly expensive. You need a compact fingerprint to check, not a full-content comparison.

  3. A content hash (or near-dup signature like SimHash) checked against seen Content Hashes catches mirrors and boilerplate that are different URLs but the same page — saving storage and indexing effort.

Write raw pages to a Content Store (blob storage). Compute a content fingerprint (a hash, or a near-dup signature like SimHash) and check it against seen Content Hashes — skip pages whose content you already have under another URL.

What the new pieces do

Content Storestore
Durable storage (blob store) for the fetched pages — the actual payload downstream indexers and analyzers consume.
Content Hashesstore
Fingerprints of page content. Catches mirror sites and boilerplate that are different URLs but the same page, so you don’t store the web twice.

Step 6 · Not all pages are equal

Prioritize the frontier

Pure BFS treats a spam page like the homepage of a major news site, and never decides when to re-crawl a page that changes hourly versus one that never changes. Crawl budget is finite; spending it evenly is wrong.

Seed URLsURL FrontierFetcher WorkersParserSeen SetPolitenessDNS CacheContent StoreContent Hashes
The system as it stands at this step. · swipe to pan the diagram

Make the frontier a priority queue: rank URLs by importance (PageRank-ish signals, depth) and schedule re-crawls by how often a page actually changes. Important, fast-changing pages get crawled sooner and more often.

Step 7 · Scale & traps

Shard out, avoid traps

One machine can’t hold the frontier or fetch the web. And the web is hostile: crawler traps (infinite calendars, endless query-param permutations) can trap a naive crawler forever in one site.

URL FrontierFetcher WorkersParserSeen SetPolitenessContent HashesSharded Frontier
New in this step: Sharded Frontier. · swipe to pan the diagram

One machine can’t hold the frontier or fetch the whole web. How do you shard it while keeping politeness?

  1. Random sharding scatters one host’s URLs across many nodes, so no single node can enforce that host’s rate limit — politeness breaks. The shard key has to keep a host together.

  2. Hashing the host to a shard puts all of a site’s URLs — and its rate limit — on one node, so per-host politeness stays local and correct. Add nodes to crawl more hosts in parallel.

  3. Full replication means every node sees every URL and they’d duplicate work or need constant coordination to divide it. You want to partition the frontier, not copy it everywhere.

Shard the frontier by host across many machines so each shard owns some domains (keeping politeness local) and the fleet scales horizontally. Defend against traps with depth limits, URL-length caps, parameter normalization and per-host page budgets.

What the new pieces do

Sharded Frontierbus
The frontier spread across many machines, partitioned by host so per-host politeness is local and the whole thing scales horizontally.

Back of the envelope

partition by host (hash)
a site’s URLs + rate limit live on one node
add nodes ⇒ more hosts in parallel
horizontal scale, politeness preserved
depth / param / length caps
defuse infinite calendars and query-param traps

You did it

You just designed a web crawler.

Seed URLsURL FrontierFetcher WorkersParserSeen SetPolitenessDNS CacheContent StoreContent HashesSharded Frontier
The finished design, end to end. · swipe to pan the diagram

Everything you assembled, in order

  • A URL frontier feeding fetch → parse → enqueue: the core crawl loop.
  • A bloom-filter Seen Set stops infinite re-crawling of the cyclic web graph.
  • Politeness: obey robots.txt and rate-limit per host so you’re not an attack.
  • A DNS cache + async resolution removes the hidden per-fetch bottleneck.
  • A content store plus content-hash dedup avoids storing mirrors and boilerplate twice.
  • A priority frontier balances importance and re-crawl freshness, not blind FIFO.
  • Shard the frontier by host to scale out, with depth/param limits to dodge traps.

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. How do you decide when to re-crawl a page?

    Estimate each page’s change rate from history (a news homepage changes hourly, an archived post never) and schedule re-crawls adaptively, often via a Poisson change model. The frontier priority blends importance and predicted freshness so crawl budget goes where it matters, not evenly.

  2. How do you handle JavaScript-rendered pages?

    A plain HTTP fetch only sees the initial HTML; SPA content needs a headless browser to render — 10–100× more expensive. So you don’t render everything: detect JS-dependent pages and route only those to a smaller headless-render fleet, keeping the cheap path for the static majority.

  3. What exactly is a crawler trap and how do you escape?

    A trap generates effectively infinite URLs — an endless calendar (?date=…), faceted-search permutations, or session ids minting a new URL each visit. Defenses: depth limits, URL-length and parameter-count caps, parameter normalization, per-host page budgets, and detecting low-value near-duplicate explosions.

  4. Bloom-filter false positives skip real pages — acceptable?

    Yes, by design: a false positive means a genuinely new URL is occasionally treated as "seen" and skipped — you miss a page. A false negative (re-crawling) never happens. Missing a tiny fraction of pages is a fine trade for the memory savings; back the filter with a durable exact set to confirm hits if needed.

  5. How do you keep fetchers busy under per-host rate limits?

    Breadth across hosts. Because politeness is per-host, the frontier interleaves URLs from thousands of domains so each host is fetched gently while the fleet stays saturated. Connection reuse (keep-alive) and async I/O let one worker handle many in-flight fetches instead of blocking on a slow host.

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. The core of a crawler is…

    • A recursive function
    • A URL frontier with a fetch→parse→enqueue loop
    • A single database

    The frontier is the to-visit queue; decoupling fetch (network) from parse (CPU) lets each scale.

  2. A bloom-filter Seen Set exists to…

    • Store page content
    • Stop infinite re-crawling of the cyclic web, compactly
    • Speed up DNS

    O(1) "probably seen" in a few bits per URL — may skip a new URL, never re-crawls a seen one.

  3. Politeness (robots.txt + rate limits) is enforced…

    • Globally across the crawler
    • Per host
    • Per worker

    It’s a per-domain promise — crawl many hosts in parallel, gently on each.

  4. URL dedup and content dedup are…

    • The same thing
    • Different jobs — a new URL can still return a stored page
    • Both done by DNS

    The bloom filter dedups URLs; content fingerprints catch mirrors/boilerplate under new URLs.

  5. The frontier is sharded by host so that…

    • URLs sort alphabetically
    • A host’s URLs and rate limit stay on one node
    • Pages compress better

    Partition-by-host keeps per-host politeness local while scaling out across machines.

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.

  • The loop: from seed URLs, fetch a page, extract its links, enqueue the new ones — forever.
  • Don’t loop: never re-crawl a URL you’ve already queued on the cyclic web graph.
  • Be polite: obey each site’s robots.txt and cap requests per host.
  • Store pages: persist fetched HTML for downstream indexers, without keeping duplicates.
  • Prioritize: crawl important, fast-changing pages sooner and re-crawl by freshness.

The qualities that shape everything

Each one names the mechanism that buys it.

Scale to billions of pages
Shard the frontier by host across many machines and run a large fetcher fleet — fetch (network-bound) and parse (CPU-bound) decoupled so each scales on its own.
Never loop forever on the cyclic graph
A bloom-filter Seen Set answers “seen this?” in O(1) at a few bits per URL — it may rarely skip a new URL but never re-crawls a seen one.
Don’t look like an attack (and get banned)
Per-host rate limits and robots.txt, crawling thousands of hosts in parallel — gentle on each, fast in aggregate.
DNS is not the hidden bottleneck
A DNS cache keyed by host with a TTL, resolved asynchronously; most pages on a site share one IP, so the first lookup serves thousands of fetches.
Don’t store the web twice
Fingerprint page content (a hash or SimHash) and skip mirrors, print versions and session-ID URLs that differ only in the URL.
Survive hostile pages
Depth limits, URL-length and parameter caps, and per-host page budgets defuse crawler traps like infinite calendars and query-param explosions.

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.

Bloom-filter Seen Set over an exact hash set

An exact set of billions of URLs is hundreds of GB — too big for memory per node. A bloom filter is “mostly right and tiny”: a rare false positive skips a new URL, but it never re-crawls a seen one.

Per-host politeness over a global request cap

A global cap still lets all that traffic pile onto one unlucky small host. Politeness is a per-domain promise — pace each host, run many hosts in parallel.

Content fingerprint over URL dedup alone

The Seen Set dedups URLs, not content — a brand-new URL can still return a page you’ve stored a thousand times. A content hash catches mirrors and boilerplate the URL check can’t.

Priority frontier over blind FIFO / BFS

Pure BFS treats a spam page like a major homepage and never decides when to re-crawl. Encoding importance and change-rate into frontier priority is what separates a toy from a search-engine crawler.

Shard by host over shard by URL

Random URL sharding scatters one host’s URLs across nodes, so no node can enforce that host’s rate limit — politeness breaks. Hashing the host keeps a site’s URLs and its rate limit together on one node.

What this teaches

Learn system design by building a distributed web crawler step by step. An interactive guide covering the URL frontier, the fetch–parse–enqueue loop, dedup with bloom filters, politeness and robots.txt, DNS caching, content storage and dedup, and sharding the frontier.

Key takeaways

  • A URL frontier feeding fetch → parse → enqueue: the core crawl loop.
  • A bloom-filter Seen Set stops infinite re-crawling of the cyclic web graph.
  • Politeness: obey robots.txt and rate-limit per host so you’re not an attack.
  • A DNS cache + async resolution removes the hidden per-fetch bottleneck.
  • A content store plus content-hash dedup avoids storing mirrors and boilerplate twice.
  • A priority frontier balances importance and re-crawl freshness, not blind FIFO.
  • Shard the frontier by host to scale out, with depth/param limits to dodge traps.

Concepts covered

  • What is a web crawler?
  • Fetch, parse, enqueue
  • Have I seen this URL?
  • Politeness & robots.txt
  • DNS at scale
  • Store & dedup content
  • Prioritize the frontier
  • Shard out, avoid traps
RUN IT YOURSELF

The "seen URLs" Bloom filter

A crawler must skip URLs it has already visited, but a billion URLs is a lot of memory. A Bloom filter answers "seen?" in a tiny bit-array. Here it is in both languages, running live. Switch tabs, read the comments, and hit Run.

HOW TO READ THE CODE — 4 IDEAS
  1. A Bloom filter is a bit-array plus k hash functions — no URLs stored.
  2. To add a URL, set the k bits its hashes point to (step 2).
  3. "Seen?" is true only if all k bits are already set (step 3).
  4. It has no false negatives (a real "seen" is never missed) but a tunable false-positive rate.
CPython · WebAssembly
built to be crawled, not memorized — make the calls, drop the seen-set, 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