Scaling Memcache at Facebook (paper)

Ideas you will meet

Caching

Cache-aside, write-through, eviction and invalidation strategies.

Distributed Systems

Consensus, time, failure and the fundamentals of scale-out.

What we already know about this

Our own lessons on these ideas. Read them here before you start, so it makes sense.

• System Design Foundations · 11 minCachingHow keeping popular data in fast memory makes systems quicker, and the invalidation problems that come with it.Read hereClose

  Why a cache helps

  Reading from memory is thousands of times faster than a database round trip (lesson 3). A cache keeps copies of frequently needed data in fast storage so most requests never reach the slower source. The results: lower latency, less load on the database, and lower cost.

  100%

  Figure SDA-FND-06-1 · Cache-aside read path · © forcsguy.com

  Before: every read hits the database

  100%

  After: a cache answers most reads

  100%

  What changed

  • Most reads never reach the database
  • Responses are faster because memory is quicker than disk
  • The cache can be stale, so you decide how long is acceptable

  Where caches live

  • In the browser: controlled with HTTP headers such as Cache-Control.
  • At the edge (CDN): copies of pages and files near the user (lesson 9).
  • In the application tier: a shared cache like Redis or Memcached.
  • Inside the database: buffer pools and query caches.

  You can use several at once, each answering whatever it can.

  Common strategies

  Strategy	How it works	Good for
  Cache-aside	App checks cache, on miss reads the database and fills the cache	the default for most read-heavy data
  Read-through	The cache itself loads from the database on a miss	simpler app code
  Write-through	Writes go to the cache and the database together	data you read right after writing
  Write-behind	Writes go to the cache first and reach the database later	very high write rates, if you can risk losing a little

  Write-behind is fast but dangerous: if the cache fails before it flushes, writes are lost.

  Writes can use the cache in three different ways. The sequence shows when the database learns about the change:

  100%

  Figure SDA-FND-06-2 · Write-through: update the cache and the database together · © forcsguy.com

  100%

  Figure SDA-FND-06-3 · Write-back: acknowledge first, save later · © forcsguy.com

  100%

  Figure SDA-FND-06-4 · Write-around: skip the cache on writes · © forcsguy.com

  Watch one read with and without a warm cache. Press Play both together, and use the Simulate tab to slow the network and see the gap grow.

  Before: no cache

  100%

  After: cache hit

  100%

  What changed

  • The database is not asked at all
  • Two network hops disappear from the request
  • The saving grows as the network gets slower

  The hard part: invalidation

  There are only two hard things in computer science: cache invalidation and naming things.

  A cache holds a copy, and copies go stale. Your options:

  1. Time to live (TTL): entries expire after a set time. Simple, but data can be stale until then.
  2. Explicit invalidation: delete or update the cache entry when the source changes. Fresher, but you must never forget a code path.
  3. Versioned keys: put a version in the key (product:42:v7) so a change simply starts using a new key.

  Most real systems combine a TTL (a safety net) with explicit invalidation on writes.

  100%

  Figure SDA-FND-06-5 · Defending the database from lookups of keys that do not exist · © forcsguy.com

  Eviction: what to drop when full

  Memory is limited, so a cache removes items when it fills. Common policies: LRU (least recently used), LFU (least frequently used) and FIFO. LRU is a good default.

  Failure modes to know

  • Cache stampede: a hot key expires and many requests rebuild it at once. Fix with a lock, request coalescing, or adding random jitter to expiry times.
  • Cache penetration: requests for keys that do not exist always miss and reach the database. Fix by caching "not found" briefly, or with a Bloom filter.
  • Cold start: after a restart the cache is empty and the database takes the full load. Warm the cache or ramp traffic up.
  • Inconsistency: two writers update the cache and database in different orders. Prefer deleting the cache entry on write over updating it.

  A worked example: what a cache is really worth

  Suppose a product page reads from a database in 20 ms and from a cache in 1 ms, and the site serves 10,000 reads per second.

  Cache hit rate	Average latency	Reads reaching the database
  0% (no cache)	20 ms	10,000 per second
  50%	10.5 ms	5,000 per second
  90%	2.9 ms	1,000 per second
  99%	1.2 ms	100 per second

  The average latency is hit x 1 + (1 - hit) x 20. Notice two things. First, going from 90% to 99% barely changes the average (2.9 to 1.2 ms) but cuts database load ten times. That is usually the real prize: a cache is a way to protect the database. Second, the first 50% is cheap and the last 9% is expensive: it needs a much bigger cache, because the remaining misses are the long tail of rarely requested items.

  How big should it be? Traffic is usually skewed: a small share of items gets most requests (often described as "80% of requests go to 20% of items", and for popular content it is more extreme). Estimate your working set: the items requested within your cache lifetime. If 2 million product pages exist but 100,000 get 90% of the traffic, a cache that holds 150,000 of them at about 5 KB each needs only 750 MB. Measure the hit rate while changing the size, and stop growing when the curve flattens.

  100%

  Figure SDA-FND-06-6 · Hit rate rises quickly, then flattens · © forcsguy.com

  Stampedes: when a popular key expires

  Take a homepage fragment read 5,000 times a second and cached for 60 seconds. At the moment it expires, thousands of requests miss at once, and every one of them rebuilds it from the database. The database, which was lightly loaded because of the cache, suddenly takes a spike. If the rebuild is slow, requests pile up, the spike lasts longer, and the system can fall over exactly when it is busiest.

  Ways to defend:

  • Request coalescing (single flight). Let one request rebuild the value while the others wait for its result.
  • A short lock on rebuild. The first request takes a lock and rebuilds; the rest either wait briefly or serve the old value.
  • Stale-while-revalidate. Keep serving the expired value for a few more seconds while one background request refreshes it. Users see slightly old data and never see a spike.
  • Jittered expiry. Add a random amount to each TTL (for example 60 seconds plus up to 10) so thousands of keys written together do not all expire together.
  • Refresh ahead. For a handful of known-hot keys, refresh them on a schedule before they expire.

  100%

  Figure SDA-FND-06-7 · One request rebuilds, the others wait or use the old value · © forcsguy.com

  Mistakes people make

  • Caching without a plan for invalidation. "Cache it for an hour" is a promise that users may see hour-old data. Decide what staleness each kind of data tolerates (a like count: minutes; a price or permission: seconds or never).
  • Caching per-user data under a shared key. If a personalised page is cached under a key that does not include the user, the next visitor sees someone else's data. Include everything that changes the response in the key, or do not cache it.
  • Treating the cache as the source of truth. Caches lose data on restart or eviction. The database must be able to rebuild everything.
  • Ignoring the hot key. One enormously popular key can overload a single cache node. Replicate that key across nodes or add a small in-process cache in front.
  • Not measuring. Track hit rate, evictions, memory use and the latency of misses. A cache you do not measure is a guess.

  Try it yourself

  In the System Builder, open "The database becomes the bottleneck". Run it without a cache and watch the database turn red, then add a cache and try different sizes. In the sandbox, raise the "popular-item skew" slider and notice how much a small cache helps when a few items dominate. You can also open the cached design on the free canvas and try wiring the cache in front of different parts.

  More on this site

  Related reading

  • Flannel: An Application-Level Edge Cache to Make Slack ScaleSlack Engineering · SlackCase Studies
  • Redis UniversityRedis · Redis Ltd.Courses
  • Redis BlogRedis engineers · Redis Ltd.Blogs
  • RedisRedis · Redis Ltd.GitHub

  Show 6 moreShow fewer

  • Redis DocumentationRedis · Redis Ltd.Documentation
  • TAO: Facebook's Distributed Data Store for the Social Graph (paper)Bronson et al. · Meta (Facebook)Papers
  • Build Your Own XCodeCrafters · GitHubGitHub
  • Memory Optimization in Redis at WattpadWattpad engineers · WattpadCase Studies
  • Redis Fleet at HerokuHeroku engineers · HerokuCase Studies
  • Redis at TrivagoTrivago engineers · TrivagoCase Studies

  Open this lesson with its quiz and exercise

• System Design Foundations · 14 minPutting it together: design a URL shortenerA complete worked design using everything so far: requirements, estimates, data model, key generation, caching, scaling and failure handling.Read hereClose

  This lesson uses every earlier idea on one concrete problem. Read it as a model of how to talk through a design.

  1. Clarify requirements

  Functional

  • Given a long URL, create a short one (sho.rt/Ab3dE9).
  • Opening the short URL redirects to the long one.
  • Optional custom alias and expiry (we will add these later).

  Non-functional

  • Redirects are fast (under about 100 ms) and highly available. If redirects fail, every link on the internet that points to us is broken.
  • Short keys must be unique. Links must not disappear unexpectedly.
  • Keys should not be guessable in sequence (so people cannot enumerate every link).

  Out of scope for now: user accounts, detailed analytics.

  2. Estimate

  Using the numbers from lesson 3: 100 million new links a month is about 40 writes per second. With reads at 10 times writes, about 400 reads per second, peaking maybe at 2,000. Five years of data is about 6 billion records at roughly 500 bytes each, so around 3 TB.

  Conclusion: traffic is modest and data is small enough for a single replicated relational database for a long time. The design should be simple, with the focus on the read path.

  3. High-level design

  100%

  Figure SDA-FND-12-1 · URL shortener architecture · © forcsguy.com

  Create a link (write path)

  1. Client sends POST /links with the long URL.
  2. The server obtains a unique ID and encodes it into a short key.
  3. It stores (key, long_url, created_at, expires_at) in the database.
  4. It returns the short URL.

  Open a link (read path)

  1. Client requests GET /Ab3dE9.
  2. The server looks in the cache. On a hit it returns the redirect.
  3. On a miss it reads the database, fills the cache, then redirects.

  4. The interesting parts

  Generating short keys

  Use base62 (a-z, A-Z, 0-9). Seven characters give 62^7, about 3.5 trillion keys, far more than the 6 billion we need.

  Options for producing a unique key:

  Approach	Good	Bad
  Hash the URL and take 7 characters	stateless	collisions, same URL gives same key
  Random key, check the database	simple	needs a check and retry, extra write load
  Unique ID, then base62	no collisions by construction	needs an ID generator, and sequential IDs are guessable

  A good choice: obtain IDs from a generator (a counter handed out in ranges to each server, or a Snowflake-style 64-bit ID), then base62-encode. To stop people guessing the next link, scramble the ID with a reversible function before encoding.

  100%

  Figure SDA-FND-12-2 · Two ways to make a short key · © forcsguy.com

  How base-62 conversion works: divide the number by 62 repeatedly and read the remainders from the last to the first. The number 11157 gives remainders 59, 55 and 2, and those map to the symbols X, T and 2, so the key is 2TX. With seven positions there are 62^7, about 3.5 trillion, possible keys.

  The database

  One table is enough:

  Column	Notes
  key (primary key)	the short code
  long_url	the target
  created_at, expires_at	for expiry

  Lookup is always by key, so the primary key index serves every read. A relational database works; a key-value store would too. Start with the one you can operate.

  The read path and the cache

  The same few links are opened over and over. A cache such as Redis in front of the database absorbs most reads. With cache-aside and a TTL of a day, plus LRU eviction, the hit rate is typically very high. Redirect responses are tiny, so the cache is small.

  A 301 redirect (permanent) can be cached by browsers, reducing traffic but hiding repeat clicks from analytics. A 302 (temporary) keeps every click visible. Choose based on whether you need accurate click counts.

  The two flows end to end

  100%

  Figure SDA-FND-12-3 · Creating a short link · © forcsguy.com

  100%

  Figure SDA-FND-12-4 · Following a short link · © forcsguy.com

  5. Scale and failure

  • Application servers are stateless, so add more behind the balancer (lesson 2).
  • Database reads: add read replicas if the cache is not enough. Writes are only about 40 per second, so sharding is unnecessary. If it ever becomes necessary, shard by key hash (lesson 5).
  • Cache failure: the database must survive a cold cache. Warm it, or ramp traffic back gradually.
  • Database failure: replicas with automatic failover, backups tested by restoring.
  • Abuse: rate limit link creation per client, and check submitted URLs for malware, since shorteners are popular with spammers.
  • Unknown keys: return 404 and cache the miss briefly, so repeated lookups of missing keys do not reach the database (lesson 6).

  6. What would you change at 100 times the size?

  Trace the bottleneck honestly: first the cache and replica count, then the ID generator becoming a coordination point, then the database size. At that point you would shard by key hash, use a multi-region deployment with a global key space, and move click counting fully onto an event stream (lesson 8).

  Worked numbers, end to end

  Assume 100 million new short links a month and a read to write ratio of 100 to 1.

  • Writes: 100,000,000 / (30 x 86,400) = about 40 per second on average, perhaps 120 at peak.
  • Reads: 40 x 100 = 4,000 per second on average, 12,000 at peak.
  • Storage: about 500 bytes per record (long URL, short key, dates, owner). Over 5 years: 100M x 12 x 5 = 6 billion links x 500 bytes = 3 TB. That fits on a handful of database nodes, so the data size is not the hard part.
  • Key length: with base 62 (a to z, A to Z, 0 to 9), 7 characters give 62^7 = about 3.5 trillion keys, comfortably more than the 6 billion we need, with room to spare against guessing.
  • Cache: if 20% of links get 80% of reads, caching the hottest 20% of one day's reads needs roughly 4,000 x 86,400 x 0.2 x 500 bytes = about 35 GB, which one or two cache nodes hold.

  So the real engineering problems are not size. They are generating unique keys without collisions at scale, keeping redirects fast, and not going down when one link goes viral.

  100%

  Figure SDA-FND-12-5 · Capacity numbers for the shortener · © forcsguy.com

  Redirect: 301 or 302?

  A short link answers with a redirect. 301 (permanent) lets browsers and intermediaries cache the redirect, which cuts load on your servers but means you never see the later clicks, so click analytics become inaccurate, and you cannot change the destination for people who cached it. 302 (temporary) is not cached, so every click reaches you: more traffic, but accurate analytics and the ability to change or disable a link immediately. Products that sell analytics, or must be able to remove abusive links, use 302 (or 307). Mention the trade-off in an interview; it shows you think about product consequences.

  Handling a link that goes viral

  One link suddenly gets 50,000 requests per second. Everything funnels onto one key: one cache node, one database row. Defences, in order of cost:

  1. Cache the redirect in a CDN with a short TTL (say 60 seconds). Most of those requests never reach you.
  2. Replicate hot keys across several cache nodes, or keep a tiny in-memory cache on each app server for the hottest few hundred keys.
  3. Rate limit per client so a script cannot add to the storm.
  4. Degrade gracefully: if the database is struggling, serve from cache even past TTL.

  100%

  Figure SDA-FND-12-6 · Layers of defence for a viral link · © forcsguy.com

  Mistakes people make

  • Guessable keys. Sequential ids let anyone enumerate every link. Use random or encoded ids if links may be private.
  • Collisions from hashing the URL. Truncating a hash of the long URL will eventually collide. Detect and retry, or generate keys from a counter or a pre-generated pool.
  • Forgetting abuse. Short links hide destinations, so they are used for phishing. Check destinations against blocklists and let people report links.
  • A single counter as a single point of failure. If one service hands out every id, it needs replicas or ranges handed to each app server in blocks.
  • Ignoring expiry and cleanup. Decide what happens to unused or expired links; otherwise storage and the key space fill with dead entries.

  Try it yourself

  Use the System Builder challenges "The database becomes the bottleneck" and "A viral link", which follow this shortener's growth. Then build the Snowflake ID lab and think about whether a time-based id is suitable for short keys. To build your own, open the URL shortener on the free canvas.

  More on this site

  Related reading

  • MongoDB: Shard KeysMongoDB · MongoDB Inc.Documentation
  • Consistent Hashing and Random Trees (paper)Karger et al. · MITPapers
  • TAO: Facebook's Distributed Data Store for the Social Graph (paper)Bronson et al. · Meta (Facebook)Papers
  • Hello Interview: System Design in a HurryHello InterviewCourses

  Show 6 moreShow fewer

  • Hello Interview: System Design WalkthroughsEvan King & Stefan Mai · Hello InterviewYouTube
  • The System Design PrimerDonne Martin · donnemartinGitHub
  • TAO: The power of the graphMeta Engineering · MetaCase Studies
  • Megastore: Providing Scalable, Highly Available Storage (paper)Baker et al. · GooglePapers
  • Sharding MySQL at TwilioTwilio engineers · TwilioCase Studies
  • Sharding out Elasticsearch at VintedVinted engineers · VintedCase Studies

  Open this lesson with its quiz and exercise

• Classic System Designs · 12 minDesign a news feedShow each user a fast, personalised list of posts from people they follow: fan-out on write versus on read, and the hybrid that works at scale.Read hereClose

  The problem

  A news feed shows a user a stream of recent posts from the accounts they follow, newest or most relevant first. It must load fast, even for users who follow thousands of accounts, and new posts should appear within seconds.

  Two operations

  • Publish a post: store it and make it visible to followers.
  • Get feed: return the first page of the user's feed quickly.

  Estimates

  Say 300 million daily users who open the feed several times a day. Reads vastly outnumber posts, so read speed is the priority. Feeds are hot data, which suits caching.

  Two basic strategies

  Fan-out on write (push)

  When a post is published, immediately push its id into each follower's feed list (kept in a cache such as Redis).

  • Reading a feed is a single quick lookup.
  • Publishing is expensive for users with many followers, and wasted for followers who never come back.

  Fan-out on read (pull)

  Build the feed when it is requested: fetch recent posts of everyone the user follows and merge.

  • Publishing is cheap.
  • Reading is slow and heavy, especially for users who follow many accounts.

  Before: fan-out on read

  100%

  After: fan-out on write

  100%

  What changed

  • Opening the feed becomes one cheap read
  • Posting costs more because it writes to every follower
  • Accounts with millions of followers need the pull approach

  The hybrid that works

  Use push for most accounts and pull for very popular ones.

  100%

  Figure SDA-CSD-07-1 · Hybrid feed: push for most, pull for popular accounts · © forcsguy.com

  Publish flow

  1. Save the post.
  2. Put a job on a queue so the request returns quickly.
  3. Workers look up the author's followers (from a graph or follower store) and push the post id to each follower's feed list, keeping only the latest few hundred.
  4. For accounts above a follower threshold, skip step 3.

  Read flow

  1. Fetch the user's feed list (post ids) from the feed cache.
  2. Fetch recent posts from followed popular accounts and merge.
  3. Hydrate the ids by loading post content, author profile and like counts from caches, falling back to the database.
  4. Rank and return a page with a cursor for the next one.

  Storing only ids in feeds keeps them small. Content lives once, in the post store and its cache.

  The two flows

  100%

  Figure SDA-CSD-07-2 · Publishing a post · © forcsguy.com

  100%

  Figure SDA-CSD-07-3 · Reading the feed · © forcsguy.com

  Components

  Component	Role
  Post service and store	Create and store posts
  Social graph	Who follows whom
  Fan-out workers	Build feeds asynchronously
  Feed cache	Per-user list of recent post ids
  Content and counter caches	Fast hydration
  Ranking	Order by recency, relevance and relationships

  The cache layers

  A feed system leans on separate caches for separate jobs: the feed lists (post ids per user, capped to a few hundred or thousand since few users scroll further), the content of posts (with a hot tier for popular ones), the social graph (who follows whom, often held in a graph-friendly store), actions (did this user like or reply to this post) and counters (likes, replies, followers). Fan-out also respects settings: a muted or hidden account is filtered out when feeds are built, even though the follow still exists. Media files go through a CDN rather than the feed service.

  Other concerns

  • Pagination: use cursors (the id or time of the last item), not page numbers, so new posts do not shift pages.
  • Deletes and privacy changes: mark posts as deleted and filter at read time, since removing ids from millions of feeds is expensive.
  • Cold start: a new or returning user has an empty feed, so build it by pull the first time.
  • Consistency: feeds are eventually consistent. A post appearing a few seconds late is acceptable.
  • Hot keys: viral posts concentrate reads, so cache and replicate them widely.

  Worked numbers: why a hybrid is needed

  Take 300 million daily users, each opening the feed 5 times a day, and an average user with 300 followers.

  • Reads: 300M x 5 = 1.5 billion feed reads a day, about 17,000 per second, 50,000 at peak. Feeds are read far more than they are written, which favours doing work at write time.
  • Fan-out on write for an ordinary user: one post becomes 300 feed insertions. If 10 million posts a day are made, that is 3 billion insertions a day, about 35,000 per second. Heavy, but steady and it runs in the background.
  • Fan-out on write for a celebrity with 50 million followers: one post becomes 50 million insertions. At 100,000 insertions per second that takes 500 seconds, over eight minutes, during which the post appears to some followers and not others, and the write workers are busy with nothing else. Posting twice an hour would keep the system saturated.
  • Fan-out on read for the same celebrity: nothing happens at write time. Each follower's feed request pulls the celebrity's latest posts and merges them: cheap per request, because one celebrity's recent posts are tiny and sit in cache.

  Hence the hybrid: push for ordinary accounts (fast reads, bounded write cost), pull for very popular ones (no write explosion), merged when the feed is built.

  100%

  Figure SDA-CSD-07-4 · The cost of one post, push versus pull · © forcsguy.com

  Ranking: chronological versus relevance

  A feed can be ordered by time (simple, predictable) or by predicted relevance (better engagement, much more machinery). A relevance feed is usually built in stages: candidate generation gathers a few hundred recent posts from the people you follow; ranking scores each with a model that uses features such as who posted, how you interacted before, post age and content type; filtering removes seen, blocked or policy-violating posts; and the top results are stored as the feed page. Mention that ranking changes the cache design: ranked pages are personal and short-lived, whereas the lists of post ids are shared and reusable.

  100%

  Figure SDA-CSD-07-5 · A ranked feed in stages · © forcsguy.com

  Mistakes people make

  • Offset paging on a live feed. New posts shift the list, so users see duplicates or miss items. Use a cursor ("posts older than this id"); see the keyset pagination lab.
  • Storing full posts in every follower's feed. Store post ids and fetch the content from a shared cache; otherwise an edit or delete must be repeated millions of times.
  • Forgetting deletes and blocks. A deleted or blocked post must disappear from feeds that already contain it; filter at read time.
  • Treating all users alike. Inactive users do not need their feeds precomputed. Build lazily on their return.
  • No limit on feed length. Keep only the latest few hundred ids per user and fall back to a slower path for older history.

  Try it yourself

  The write-heavy System Builder challenge "A burst of writes" shows the same pressure as a celebrity post: many writes that the database cannot apply at once. Try it with and without a queue, and compare how it feels to push work into the background. To model the feed's write path yourself, open the social feed on the free canvas.

  More on this site

  Related reading

  • TAO: Facebook's Distributed Data Store for the Social Graph (paper)Bronson et al. · Meta (Facebook)Papers
  • TAO: The power of the graphMeta Engineering · MetaCase Studies
  • Solving Remote Build Cache Misses (2 parts) at SoundCloudSoundCloud engineers · SoundCloudCase Studies
  • How we built rate limiting capable of scaling to millions of domainsCloudflare engineering · CloudflareCase Studies

  Show 6 moreShow fewer

  • Meta Engineering BlogMeta Engineering · MetaBlogs
  • Netflix Technology BlogNetflix Engineering · NetflixBlogs
  • Scaling Memcache at Facebook (paper)Nishtala et al. · Meta (Facebook)Papers
  • Backend Developer Roadmaproadmap.sh community · roadmap.shCourses
  • Facebook AI Research Engineering BlogFacebook AI Research engineers · Facebook AI ResearchBlogs
  • Latest News from Google Research BlogGoogle engineers · GoogleBlogs

  Open this lesson with its quiz and exercise