79:T48c,Before diving into technical requirements, we must understand Instagram's product landscape. Instagram is no longer just a photo app—it's a multi-modal content platform encompassing several distinct but interconnected features: **Core Content Types:** - **Feed Posts:** Permanent photos and videos shared to a user's profile and followers' feeds - **Stories:** Ephemeral 24-hour content with rich interactive features - **Reels:** Short-form vertical videos competing with TikTok - **Direct Messages:** Private photo/video sharing with end-to-end encryption - **IGTV/Long-form Video:** Extended video content for creators - **Live Streaming:** Real-time video broadcasts with viewer interaction **Discovery Surfaces:** - **Home Feed:** Chronologically-influenced algorithmic content from followed accounts - **Explore Page:** Personalized discovery of content from unfollowed accounts - **Reels Tab:** Dedicated short-form video discovery feed - **Search & Tags:** Hashtag exploration and account discovery **Engagement Features:** - Likes, comments, shares, and saves - Story reactions and replies - Direct message interactions - Collaborative posts and tags7a:T422,The Explore page is Instagram's content discovery engine—presenting personalized content from accounts the user doesn't yet follow. This is distinct from the Home Feed (which shows followed accounts) and represents Instagram's investment in **recommendation** rather than mere **distribution**. **Explore Objectives:** 1. **Engagement maximization:** Surface content users will interact with (like, comment, save, share) 2. **Session extension:** Keep users browsing longer through compelling content cascades 3. **Interest expansion:** Introduce users to new topics/accounts they'll enjoy 4. **Creator discovery:** Connect emerging creators with potential audiences 5. **Advertiser integration:** Blend promoted content seamlessly with organic recommendations **What Makes Explore Different:** Unlike the Home Feed where content comes from a known set (followed accounts), Explore must select from **all public content on Instagram**—billions of posts. This is a fundamentally harder problem requiring sophisticated candidate generation and ranking.71:{"json":{"id":"seed_86bd98a57f9aee28330131947322a8c1","slug":"instagram-photos","category":"System Design (HLD)","topic":"Instagram Photos","title":"Designing Instagram: Photo-Sharing at Planetary Scale","description":"Master the architecture behind Instagram's photo-sharing platform. Learn how to design systems for photo upload, processing, feed generation, Stories, Explore recommendations, and scaling media storage to handle billions of daily interactions across 2+ billion users.","difficulty":"ADVANCED","estimatedMinutes":120,"pages":[{"id":"seed_c5e22f0cbf03672c2e0ec3a55f2bb6db","title":"Requirements Analysis: Upload, Share, Explore","order":0,"content":[{"id":"opening-hook","title":"The Photo-Sharing Revolution","content":[{"data":{"content":"Instagram fundamentally transformed how humanity captures and shares visual moments. What began as a simple photo-filtering application has evolved into a comprehensive visual communication platform with **over 2 billion monthly active users** sharing **more than 2 billion photos and videos daily**. Behind every double-tap, every story swipe, and every Explore scroll lies a sophisticated distributed system that must handle astronomical scale while maintaining the instantaneous, seamless experience users expect.\n\nDesigning Instagram isn't merely a photo storage problem—it's a complex orchestration of **real-time media processing**, **personalized content delivery**, **social graph traversal**, and **recommendation algorithms**, all operating under extreme latency constraints. A 100-millisecond delay in feed loading can measurably impact user engagement. A processing failure during the upload of a viral moment could mean missing the cultural zeitgeist entirely."},"type":"text"},{"data":{"icon":"Target","title":"What You Will Learn","content":"By the end of this page, you will have a comprehensive understanding of Instagram's requirements—both functional capabilities the system must provide and non-functional constraints it must satisfy. You will learn to decompose a seemingly simple 'photo-sharing app' into its constituent technical challenges, establishing the foundation for architectural decisions in subsequent pages.","variant":"info"},"type":"callout"}]},{"id":"instagram-ecosystem","title":"Understanding the Instagram Ecosystem","content":[{"data":{"content":"$79"},"type":"text"},{"data":{"icon":"Info","title":"Scope Decision for This Design","content":"For this system design, we will focus on the core photo-sharing functionality: **Feed Posts, Stories, Feed Generation, and Explore recommendations**. While Instagram also includes video, messaging, and live features, the photo-centric components represent the foundational architecture that other features build upon. Understanding photo handling thoroughly provides the conceptual framework for all media types.","variant":"note"},"type":"callout"},{"data":{"content":"**The Scale We're Designing For:**\n\nInstagram's scale is genuinely planetary. Consider these order-of-magnitude figures:"},"type":"text"},{"data":{"rows":[["Monthly Active Users","~2 billion"],["Daily Active Users","~500 million"],["Photos shared daily","~2 billion"],["Stories posted daily","~500 million"],["Feed views per day","~10+ billion"],["Average photos per user","~200"],["Peak concurrent users","~100+ million"]],"headers":["Metric","Scale"]},"type":"table"},{"data":{"content":"These numbers aren't abstract—they directly inform every architectural decision. A design that works for 1 million users may catastrophically fail at 1 billion. Understanding scale is the first step toward designing for it."},"type":"text"}],"titleType":"number","titleNumber":1},{"id":"functional-requirements-upload","title":"Functional Requirements: Photo Upload","content":[{"data":{"content":"The upload flow is the entry point for all content into Instagram. What appears simple—selecting a photo and tapping 'Share'—actually triggers a complex pipeline with numerous requirements:\n\n**Core Upload Capabilities:**"},"type":"text"},{"data":{"items":["**Multi-photo uploads:** Users can upload up to 10 photos/videos in a single carousel post","**In-app editing:** Crop, rotate, adjust brightness/contrast/saturation, apply filters","**Filter application:** Real-time preview and application of Instagram's signature visual filters","**Caption and metadata:** Text caption, location tagging, user tagging (up to 20 per post)","**Hashtag support:** Discoverable hashtags (up to 30) for content categorization","**Alt text:** Accessibility descriptions for visually impaired users","**Sharing options:** Cross-posting to Facebook, Twitter, Tumblr simultaneously","**Audience controls:** Public, private (followers-only), or close friends visibility","**Comment controls:** Turn comments on/off, filter offensive comments","**Scheduled posting:** (For business accounts) Queue content for future publication"],"title":{"icon":"Target","text":"Upload Functional Requirements","type":"icon","color":"primary","variant":"outline-circle"},"variant":"accent"},"type":"list"},{"data":{"content":"**Photo Processing Requirements:**\n\nWhen a photo enters the system, it undergoes extensive processing before becoming visible:\n\n1. **Format normalization:** Accept JPEG, PNG, HEIC, and other formats; convert to standardized storage format\n2. **Orientation correction:** Apply EXIF orientation data to correct rotated photos\n3. **Resolution scaling:** Generate multiple resolutions for different device densities (1x, 2x, 3x) and contexts (thumbnail, feed, full-screen)\n4. **Filter baking:** Permanently apply selected filter to stored versions (not applied at render time)\n5. **Compression optimization:** Balance quality vs. file size for network efficiency\n6. **EXIF stripping:** Remove sensitive metadata (GPS coordinates unless explicitly shared)\n7. **Content analysis:** Machine learning for content classification, safety scanning, accessibility features"},"type":"text"},{"data":{"rows":[["Thumbnail","150x150px","Grid view, search results, notifications","~10-15 KB per image"],["Low resolution","320x320px","Low-bandwidth preview, placeholder loading","~25-40 KB per image"],["Standard feed","640x640px to 1080x1350px","Feed display, aspect-ratio dependent","~100-200 KB per image"],["High resolution","1080x1080px to 1080x1350px","Full-screen viewing, pinch-to-zoom","~200-400 KB per image"],["Original archive","Full original resolution","Data export, backup (if retained)","1-10+ MB per image"]],"caption":"Photo Resolution Variants","headers":["Variant","Resolution","Use Case","Storage Impact"]},"type":"table"},{"data":{"icon":"Lightbulb","title":"Resolution Strategy Trade-off","content":"Instagram must balance storage costs against user experience. Storing 5+ variants per photo for 2 billion daily uploads is expensive, but computing these on-demand would create unacceptable latency. The solution is pre-generating commonly-used resolutions while dynamically generating rare variants. This 'eager common, lazy rare' pattern is fundamental to media platforms.","variant":"tip"},"type":"callout"},{"data":{"content":"**Upload State Management:**\n\nThe upload process must handle real-world conditions gracefully:\n\n- **Network interruption recovery:** Resume interrupted uploads without re-transmitting completed chunks\n- **Background uploading:** Continue upload even if user navigates away or app backgrounds\n- **Optimistic UI:** Show post in feed immediately while upload completes asynchronously\n- **Failure handling:** Retry logic with exponential backoff; queue for retry when connectivity returns\n- **Duplicate detection:** Prevent duplicate posts from double-taps or retry storms\n- **Rate limiting:** Prevent spam uploads while allowing legitimate high-volume posting"},"type":"text"}],"titleType":"number","titleNumber":2},{"id":"functional-requirements-share","title":"Functional Requirements: Content Sharing & Distribution","content":[{"data":{"content":"Once content is uploaded, it must be distributed to the appropriate audiences. 'Sharing' in Instagram's context encompasses multiple distribution mechanisms:\n\n**Feed Distribution:**\n\nWhen a user posts a photo, it must appear in their followers' feeds. This sounds trivial, but consider:\n\n- A celebrity with 500 million followers posts a photo\n- That photo must be added to 500 million individual home feeds\n- Each feed has its own algorithmic ranking for that user\n- All 500 million users should see it 'immediately'\n\nThis is the **fanout problem**—the foundational challenge of social media distribution."},"type":"text"},{"data":{"left":[{"data":{"items":["At post time, writes to all followers' feed caches","Fast reads—feed is pre-computed","Expensive writes for popular accounts","Celebrity post = millions of writes","Feed may be stale if write fails","Storage: O(followers × posts) per user"],"title":{"icon":"Zap","text":"Fanout-on-Write (Push Model)","type":"icon","color":"primary","variant":"outline-circle"},"variant":"accent"},"type":"list"}],"right":[{"data":{"items":["At read time, fetches from all followed accounts","Always fresh—query at request time","Cheap writes—single write per post","Expensive reads for prolific followers","User following 1000 accounts = 1000 queries","Storage: O(posts) per user"],"title":{"icon":"Zap","text":"Fanout-on-Read (Pull Model)","type":"icon","color":"primary","variant":"outline-circle"},"variant":"accent"},"type":"list"}]},"type":"two-column"},{"data":{"icon":"Info","title":"Instagram's Hybrid Approach","content":"Instagram uses a **hybrid fanout** strategy. Regular users use fanout-on-write (push their posts to followers' pre-computed feeds). Celebrities above a follower threshold (e.g., 10,000+) use fanout-on-read (their posts are pulled at feed-render time and merged). This hybrid balances write amplification against read latency, optimizing for the common case while handling extreme cases gracefully.","variant":"note"},"type":"callout"},{"data":{"content":"**Sharing Vectors:**"},"type":"text"},{"data":{"items":["**Public feed sharing:** Visible to all followers; appears in hashtag pages if public account","**Direct message sharing:** Private sharing to individuals or groups (up to 32 recipients)","**Story sharing:** 24-hour ephemeral sharing with view tracking","**Reshare to Stories:** Amplifying posts by sharing to your own Story","**Link sharing:** External URL for sharing outside Instagram ecosystem","**Embed sharing:** Embeddable widget code for websites and blogs","**Collaborative posts:** Share authorship and visibility across multiple accounts","**Close Friends:** Story sharing restricted to a curated list"],"title":{"icon":"Target","text":"Content Distribution Channels","type":"icon","color":"primary","variant":"outline-circle"},"variant":"glass"},"type":"list"},{"data":{"content":"**Engagement on Shared Content:**\n\nShared content generates engagement signals that feed back into the system:"},"type":"text"},{"data":{"rows":[["Like","User ID, post ID, timestamp","Updates like count, notifies author, informs ranking"],["Comment","User ID, text, timestamp, parent","Updates comment count, possibly triggers review"],["Save","User ID, collection ID","Informs content quality signals"],["Share","User ID, recipient(s), channel","Amplification signal for ranking"],["Profile visit from post","User ID, source post","Attribution for discovery funnel"],["Follow from post","User ID, followed ID, source","Conversion attribution"]],"headers":["Engagement Type","Data Generated","System Impact"]},"type":"table"},{"data":{"content":"Each engagement event becomes input to ranking algorithms, spam detection, content moderation, and business analytics. The system must capture, store, and process billions of these events daily in near-real-time."},"type":"text"}],"titleType":"number","titleNumber":3},{"id":"functional-requirements-explore","title":"Functional Requirements: Content Discovery & Explore","content":[{"data":{"content":"$7a"},"type":"text"},{"data":{"items":["**Personalization:** Content tailored to individual user interests based on historical behavior","**Freshness:** Prioritize recent content while balancing with evergreen high-quality posts","**Diversity:** Prevent filter bubbles by exposing users to varied content types and creators","**Safety:** Filter policy-violating content before presentation","**Responsiveness:** Incorporate new interest signals within hours/days, not weeks","**Cold start handling:** Provide reasonable recommendations for new users with no history","**Trending integration:** Surface viral/trending content that transcends individual interests","**Contextual awareness:** Consider time-of-day, current events, seasonal relevance"],"title":{"icon":"Sparkles","text":"Explore Recommendation Requirements","type":"icon","color":"primary","variant":"outline-circle"},"variant":"accent"},"type":"list"},{"data":{"content":"**Explore Interface Requirements:**\n\nThe Explore page is more than a list—it's a carefully designed discovery surface:\n\n- **Grid layout:** Visually scannable grid of photos/videos with varied sizing for emphasis\n- **Categories/Topics:** Filterable interest categories (Fashion, Food, Travel, etc.)\n- **Search integration:** Query-based discovery alongside algorithmic recommendations\n- **Infinite scroll:** Continuous content loading as user scrolls (no pagination)\n- **Quick preview:** Tap-and-hold to preview without full navigation\n- **Seamless transitions:** Tapping a post expands to a scrollable feed from that point\n- **Audio preview:** Muted autoplay for video content in grid view"},"type":"text"},{"data":{"icon":"AlertCircle","title":"The Explore Challenge","content":"Generating personalized recommendations from billions of candidates in <100ms is one of Instagram's hardest engineering challenges. The system must evaluate candidate generation, feature extraction, model inference, business rule application, and result assembly—all within a single page load. This requires careful architecture of candidate funnels, model optimization, and extensive caching.","variant":"warning"},"type":"callout"}],"titleType":"number","titleNumber":4},{"id":"non-functional-scale","title":"Non-Functional Requirements: Scale & Performance","content":[{"data":{"content":"Non-functional requirements define **how well** the system performs, not **what** it does. For a platform of Instagram's scale, these constraints drive most architectural decisions.\n\n**Scale Requirements:**"},"type":"text"},{"data":{"rows":[["Daily uploads","2 billion photos/videos","~25,000 uploads per second sustained, 100K+ peak"],["Feed requests","10+ billion/day","~120,000 feed renders per second"],["Storage","Exabytes of media","Requires globally distributed object storage"],["Concurrent users","100+ million peak","Massive connection handling infrastructure"],["Social graph","Billions of edges","Follower relationships, blocks, mutes"],["Event stream","Trillions of events/day","Likes, views, scrolls, impressions"],["ML inference","Millions of inferences/second","Content classification, ranking, safety"]],"caption":"Instagram Scale Metrics & Requirements","headers":["Dimension","Estimated Scale","Implication"]},"type":"table"},{"data":{"content":"**Latency Requirements:**\n\nUser experience degrades measurably with latency. Instagram's requirements reflect consumer expectations:"},"type":"text"},{"data":{"rows":[["Feed load (home)","<200ms","<500ms","First impression on app open"],["Explore load","<200ms","<500ms","Discovery engagement"],["Photo upload (client)","<3 seconds apparent","<10 seconds actual","Perceived reliability"],["Like/action response","<100ms","<300ms","Instantaneous feedback feel"],["Story load","<150ms","<400ms","Seamless swipe experience"],["Search results","<100ms","<300ms","Responsive discovery"],["Image render","<50ms from cache","<200ms from origin","Scrolling smoothness"]],"headers":["Operation","Target Latency (p50)","Target Latency (p99)","User Impact"]},"type":"table"},{"data":{"content":"**Apparent vs. Actual Latency:**\n\nNote the distinction between *apparent* and *actual* upload time. Instagram uses **optimistic UI**—the post appears in your feed immediately while upload continues in the background. This decouples perceived performance from actual processing time, a critical UX pattern for content creation apps."},"type":"text"},{"data":{"icon":"Lightbulb","title":"The P99 Matters More Than P50","content":"For user experience, the 99th percentile latency matters more than the median. If 1% of feed loads take >2 seconds, that's 10 million frustrated experiences per day. At Instagram's scale, rare events happen constantly. Design for the tail, not just the average.","variant":"tip"},"type":"callout"}],"titleType":"number","titleNumber":5},{"id":"non-functional-availability","title":"Non-Functional Requirements: Availability & Reliability","content":[{"data":{"content":"For a platform as central to social interaction as Instagram, availability is paramount. Users expect Instagram to work—always, everywhere.\n\n**Availability Targets:**"},"type":"text"},{"data":{"rows":[["Core feed/viewing","99.99% (four nines)","~52 minutes","Home feed, Stories, post viewing"],["Content creation","99.9% (three nines)","~8.7 hours","Upload, posting, commenting"],["Auxiliary features","99.5%","~1.8 days","Analytics, insights, scheduling"]],"headers":["Tier","Target","Allowed Downtime/Year","Use Cases"]},"type":"table"},{"data":{"content":"**Why Different Tiers?**\n\nNot all features require equal availability. Users tolerate brief upload failures (they'll retry) more than they tolerate being unable to view content. This tiered approach allows engineering resources to focus on what matters most."},"type":"text"},{"data":{"items":["**No data loss:** Uploaded content must never be lost (durability: 99.999999999% or 11 nines)","**Graceful degradation:** Partial failures shouldn't bring down the entire app","**Regional resilience:** Survive complete data center failures without user impact","**Disaster recovery:** Full platform recovery within hours of catastrophic failure","**Eventual consistency:** Accept temporary inconsistency for availability gains","**Idempotency:** Duplicate requests (network retries) don't cause duplicate actions","**Circuit breaking:** Failures in one service don't cascade to others"],"title":{"icon":"CheckCircle","text":"Reliability Requirements","type":"icon","color":"primary","variant":"outline-circle"},"variant":"glass"},"type":"list"},{"data":{"content":"**The CAP Theorem Reality:**\n\nInstagram's scale forces trade-offs. Under the CAP theorem (Consistency, Availability, Partition tolerance—pick two), Instagram prioritizes **Availability and Partition tolerance** over strict Consistency.\n\n**What this means in practice:**\n\n- Your like might take a few seconds to show to others (eventual consistency)\n- You might briefly see 'stale' follower counts after someone unfollows\n- A post might appear in some feeds before others during high load\n- Deletion might take time to propagate across all caches\n\nThese temporary inconsistencies are acceptable trade-offs for a platform that *must* be available. An unavailable Instagram is useless; a briefly inconsistent Instagram is merely imperfect."},"type":"text"},{"data":{"icon":"Info","title":"Consistency Boundaries","content":"While Instagram accepts eventual consistency for most operations, some require immediate consistency: authentication, account settings changes, payment processing, and content removal for policy violations. The architecture must support both consistency models, choosing appropriately per use case.","variant":"note"},"type":"callout"}],"titleType":"number","titleNumber":6},{"id":"non-functional-safety","title":"Non-Functional Requirements: Safety, Privacy & Compliance","content":[{"data":{"content":"As a global platform, Instagram must navigate complex regulatory requirements, protect user safety, and maintain trust. These requirements significantly impact architecture.\n\n**Content Safety Requirements:**"},"type":"text"},{"data":{"items":["**Pre-publication screening:** ML models scan content before it becomes visible","**Policy violation detection:** Identify nudity, violence, hate speech, misinformation","**Rapid takedown:** Remove violating content within minutes of detection","**Child safety:** CSAM detection with law enforcement reporting requirements","**Spam prevention:** Block automated spam, fake engagement, and bot networks","**Harassment protection:** Detect and filter abusive comments and messages","**Sensitive content screening:** Age-gate or blur potentially sensitive imagery","**Appeals process:** Allow users to contest moderation decisions"],"title":{"icon":"AlertCircle","text":"Safety & Moderation Requirements","type":"icon","color":"destructive","variant":"outline-circle"},"variant":"accent"},"type":"list"},{"data":{"content":"**Privacy Requirements:**"},"type":"text"},{"data":{"rows":[["GDPR compliance","EU users' data rights","Data export, deletion capabilities, consent tracking"],["CCPA compliance","California privacy rights","Similar to GDPR with CA-specific requirements"],["Data minimization","Collect only necessary data","Review of all data collection points"],["Right to be forgotten","Complete data deletion on request","Cross-service deletion propagation"],["Private account semantics","Content only visible to approved followers","Access control checks at every render"],["DM encryption","End-to-end encrypted messaging","Client-side encryption infrastructure"],["Location privacy","GPS data handling","User-controlled sharing, EXIF stripping"],["Minor protection","Enhanced privacy for under-18 users","Age detection and restriction systems"]],"headers":["Requirement","Description","Architectural Impact"]},"type":"table"},{"data":{"content":"**Compliance Architecture:**\n\nThese requirements mandate specific architectural capabilities:\n\n1. **Audit logging:** Every data access logged for compliance review\n2. **Data residency:** Ability to keep certain data in specific geographic regions\n3. **Consent management:** Tracking user consent for various data uses\n4. **Automated rights fulfillment:** Processing deletion/export requests within mandated timeframes\n5. **Age verification:** Preventing under-age access to restricted content\n6. **Cross-border transfer controls:** Compliant methods for international data movement"},"type":"text"},{"data":{"icon":"AlertCircle","title":"Safety at Scale is Hard","content":"With billions of uploads daily, even 99.9% accurate content moderation means millions of mistakes—both false positives (legitimate content blocked) and false negatives (policy-violating content published). This is why moderation combines ML (scale) with human review (accuracy) and user reporting (coverage).","variant":"warning"},"type":"callout"}],"titleType":"number","titleNumber":7},{"id":"capacity-estimation","title":"Back-of-Envelope Estimation","content":[{"data":{"content":"Let's estimate the system's capacity requirements—a critical exercise for understanding infrastructure needs and validating our design can handle the load.\n\n**Traffic Estimation:**\n\n```\nDaily Active Users (DAU): 500 million\nAverage feed views per user per day: 20\nTotal daily feed requests: 500M × 20 = 10 billion\n\nFeed requests per second (average): 10B / 86,400 ≈ 115,000 RPS\nPeak multiplier: 3x average (busy hours)\nPeak feed RPS: ~350,000 RPS\n```"},"type":"text"},{"data":{"content":"**Storage Estimation:**\n\n```\nDaily uploads: 2 billion photos\nAverage original photo size: 2 MB\nResolution variants stored: 5 (thumbnail, low, medium, high, original)\nStorage multiplier for variants: ~0.5x (smaller versions don't add much)\nTotal storage per photo: 2 MB × 1.5 = 3 MB\n\nDaily storage growth: 2B × 3 MB = 6 petabytes/day\nYearly storage growth: 6 PB × 365 = 2.2 exabytes/year\nWith 3-year retention + 3x replication: ~20+ exabytes total\n```"},"type":"text"},{"data":{"content":"**Bandwidth Estimation:**\n\n```\nDaily feed views: 10 billion\nPhotos per feed load: ~20\nAverage photo size served: 200 KB (compressed, appropriate resolution)\nData per feed load: 20 × 200 KB = 4 MB\n\nDaily egress: 10B × 4 MB = 40 petabytes/day\nPeak egress rate: 40 PB / 86,400 × 3 = ~1.4 TB/second\n```"},"type":"text"},{"data":{"rows":[["Feed QPS","115K avg, 350K peak","Horizontally scaled API fleet, aggressive caching"],["Upload QPS","~25K sustained","Dedicated upload ingestion clusters"],["Storage growth","6 PB/day","Object storage at exabyte scale (S3/GCS class)"],["Network egress","1.4 TB/s peak","Global CDN with 100+ Tbps capacity"],["Social graph","100B+ edges","Specialized graph database infrastructure"],["Event stream","10M+ events/second","Kafka/Kinesis class streaming infrastructure"]],"caption":"Capacity Summary","headers":["Dimension","Estimate","Infrastructure Implication"]},"type":"table"},{"data":{"icon":"Lightbulb","title":"Estimation as Design Input","content":"These estimates aren't just interesting numbers—they directly inform architecture. 350K QPS means we must shard aggressively. 6 PB/day means object storage (not traditional databases). 1.4 TB/s egress means CDN is not optional. Let capacity requirements drive architectural decisions, not intuition alone.","variant":"tip"},"type":"callout"}],"titleType":"number","titleNumber":8},{"id":"summary-next","title":"Requirements Summary & Architecture Preview","content":[{"data":{"content":"We've established a comprehensive understanding of Instagram's requirements. Let's consolidate what we'll be designing for:"},"type":"text"},{"data":{"type":"ordered","items":["**Photo upload** with multi-format support, editing, filtering, and metadata","**Content distribution** via hybrid fanout to followers' personalized feeds","**Stories** with ephemeral 24-hour visibility and rich interactions","**Explore discovery** with personalized recommendations from all public content","**Engagement features** including likes, comments, saves, and shares","**Privacy controls** for visibility, commenting, and message access"],"title":{"icon":"CheckCircle","text":"Functional Requirements Summary","type":"icon","color":"primary","variant":"outline-circle"},"variant":"accent"},"type":"list"},{"data":{"type":"ordered","items":["**Scale:** 2B daily uploads, 10B daily feed views, exabytes of storage","**Latency:** <200ms feed loads, <100ms interaction responses","**Availability:** 99.99% for viewing, 99.9% for creation","**Durability:** 11 nines for stored content—never lose user data","**Safety:** Pre-publication content screening, rapid takedowns","**Compliance:** GDPR, CCPA, data residency, privacy controls"],"title":{"icon":"CheckCircle","text":"Non-Functional Requirements Summary","type":"icon","color":"primary","variant":"outline-circle"},"variant":"glass"},"type":"list"},{"data":{"content":"**What's Next: The Image Processing Pipeline**\n\nWith requirements established, we'll dive into the first major component: the **image processing pipeline**. This is where uploaded photos are transformed from raw user uploads into the optimized, multi-resolution assets that power Instagram's visual experience.\n\nWe'll explore:\n\n- Upload ingestion architecture and chunked uploads\n- Asynchronous processing workflows\n- Filter application and image transformation\n- Resolution variant generation\n- Content analysis and classification\n- Storage strategies for exabyte-scale media"},"type":"text"},{"data":{"icon":"CheckCircle","title":"Requirements Complete","content":"You now have a comprehensive understanding of what we're designing: a photo-sharing platform at planetary scale. These requirements will guide every architectural decision in the pages ahead. Remember—great system design starts with clear, complete requirements. Next, we'll begin building the system that satisfies them.","variant":"success"},"type":"callout"}],"titleType":"number","titleNumber":9}],"isLocked":false},{"id":"seed_f952e9ceaebcbe18d68cebcec47a9bbe","title":"Image Processing Pipeline: From Upload to Delivery","order":1,"content":null,"isLocked":true},{"id":"seed_305d6f0c951699c5a03e324762735b4f","title":"Feed Generation: Curating the Timeline at Scale","order":2,"content":null,"isLocked":true},{"id":"seed_c3c2ab8437fe9bb21e183a269906d956","title":"Stories Architecture: Ephemeral Content at Planetary Scale","order":3,"content":null,"isLocked":true},{"id":"seed_bd4c033c7a2b699abb655fb4625069bc","title":"Explore Algorithm: Discovery at Planetary Scale","order":4,"content":null,"isLocked":true},{"id":"seed_6f062cc73c47165aa9b21aedd8731276","title":"Scaling Media Storage: Exabytes of Photos and Videos","order":5,"content":null,"isLocked":true}]}}