--- title: "How YouTube's CDN Works: Google Global Cache and Video Delivery" description: "How YouTube delivers video: Google Global Cache inside ISPs, adaptive streaming, video formats and what makes YouTube's CDN so fast." image: https://blog.blazingcdn.com/hubfs/Gemini-Blog/image-Nov-01-2026-04-25-12-2357-PM.jpeg --- [![BlazingCDN](https://blog.blazingcdn.com/hubfs/Logo/blog-logo-w.png)](https://blog.blazingcdn.com?hsLang=en-us) [๐Ÿ“˜Learn โ–พ](https://blog.blazingcdn.com/cdn-learn?hsLang=en-us) [CDN Fundamentals](https://blog.blazingcdn.com/cdn-fundamentals?hsLang=en-us) [By Content Type](https://blog.blazingcdn.com/by-content-type?hsLang=en-us) [Advanced Concepts](https://blog.blazingcdn.com/advanced-concepts?hsLang=en-us) [Glossary](https://blog.blazingcdn.com/glossary?hsLang=en-us) [โšก Web Performance](https://blog.blazingcdn.com/web-performance?hsLang=en-us) [๐ŸŽฌVideo & Streaming โ–พ](https://blog.blazingcdn.com/video-streaming-cdn?hsLang=en-us) [๐Ÿ”ด Live Streaming](https://blog.blazingcdn.com/live-streaming?hsLang=en-us) [๐Ÿ“บ VOD & OTT](https://blog.blazingcdn.com/vod-ott?hsLang=en-us) [๐Ÿ’ฐ Bandwidth & Costs](https://blog.blazingcdn.com/bandwidth-costs?hsLang=en-us) [๐Ÿญ Other Industries โ–พ](https://blog.blazingcdn.com/cdn-industry-insights?hsLang=en-us) [๐Ÿ“บ Media & Broadcasting](https://blog.blazingcdn.com/media-broadcasting?hsLang=en-us) [๐Ÿ’พ Software & SaaS](https://blog.blazingcdn.com/software-saas?hsLang=en-us) [๐Ÿ—๏ธ DevOps & Cloud Infra](https://blog.blazingcdn.com/devops-cloud-infra?hsLang=en-us) [๐Ÿ“š AdTech & Advertising](https://blog.blazingcdn.com/adtech-advertising?hsLang=en-us) [๐ŸŽฎ Gaming & Esports](https://blog.blazingcdn.com/gaming-esports?hsLang=en-us) [๐Ÿ“ฑ Mobile Apps & Developers](https://blog.blazingcdn.com/mobile-apps-developers?hsLang=en-us) [๐Ÿค– AI & Machine Learning](https://blog.blazingcdn.com/ai-machine-learning?hsLang=en-us) [๐Ÿ›๏ธ Enterprise & Corporate](https://blog.blazingcdn.com/enterprise-corporate?hsLang=en-us) [๐Ÿ“š E-Learning & EdTech](https://blog.blazingcdn.com/e-learning-edtech?hsLang=en-us) [๐ŸŸ๏ธ Sports & Live Events](https://blog.blazingcdn.com/sports-live-events?hsLang=en-us) [๐Ÿ’ฐ Pricing & Costs โ–พ](https://blog.blazingcdn.com/cdn-pricing-and-cdn-costs?hsLang=en-us) [Provider Pricing](https://blog.blazingcdn.com/provider-pricing?hsLang=en-us) [Cost Optimization](https://blog.blazingcdn.com/cost-optimization?hsLang=en-us) [Decision Support](https://blog.blazingcdn.com/decision-support?hsLang=en-us) [โšกCompare โ–พ](https://blog.blazingcdn.com/cdn-comparison?hsLang=en-us) [Provider Comparisons](https://blog.blazingcdn.com/provider-comparisons?hsLang=en-us) [Strategy Comparisons](https://blog.blazingcdn.com/strategy-comparisons?hsLang=en-us) [Ratings & Benchmarks](https://blog.blazingcdn.com/cdn-ratings-and-benchmarks?hsLang=en-us) [๐Ÿ“Š Benchmarks](https://blog.blazingcdn.com/cdn-ratings-and-benchmarks?hsLang=en-us) [๐Ÿ”’ Security โ–พ](https://blog.blazingcdn.com/cdn-security?hsLang=en-us) [Attack Protection](https://blog.blazingcdn.com/attack-protection?hsLang=en-us) [Encryption & Access](https://blog.blazingcdn.com/encryption-access?hsLang=en-us) [Video & DRM Security](https://blog.blazingcdn.com/video-drm-security?hsLang=en-us) [๐Ÿ“ Case Studies](https://blog.blazingcdn.com/case-studies?hsLang=en-us) [๐Ÿ”Œ Integrations](https://blog.blazingcdn.com/integrations?hsLang=en-us) [๐Ÿ› ๏ธ Tools](https://blog.blazingcdn.com/cdn-tools?hsLang=en-us) [Get Started](https://blazingcdn.com/sign-up-contact-form/) [Learn](https://blog.blazingcdn.com/en-us/tag/learn) [Video - Video & Streaming](https://blog.blazingcdn.com/en-us/tag/video-video-streaming) [Learn - CDN Fundamentals](https://blog.blazingcdn.com/en-us/tag/learn-cdn-fundamentals) # How YouTube's CDN Works: Google Global Cache and Video Delivery Vadym Borovskyi Nov 1, 2026, 5:27:18โ€ฏPM ![](https://blog.blazingcdn.com/hubfs/Gemini-Blog/image-Nov-01-2026-04-25-12-2357-PM.jpeg) The YouTube CDN is Google's own delivery network, and its most important tier sits inside other companies' networks. Google Global Cache (GGC) servers, installed in internet service provider (ISP) facilities, serve popular YouTube videos locally so that most bytes never cross the ISP's upstream links. YouTube has said since 2019 that more than 500 hours of video are uploaded every minute. That long-tail scale is why its CDN combines ISP-embedded caches, Google's edge, regional clusters and demand-based encoding. This article covers the request flow, the VP9 and AV1 codec strategy, a worked ISP offload calculation, a comparison with Netflix Open Connect, and the failure modes worth instrumenting on your own platform. ## **YouTube CDN architecture: four tiers from ISP rack to origin** YouTube delivery is a hierarchy where each tier exists to keep requests away from the tier above it. Distance matters, but delivery cost and link congestion matter more: a byte served from inside the viewer's ISP costs neither side anything at an interconnect. - **Google Global Cache nodes:** Google-owned servers inside ISP networks. They hold the content that is popular in that ISP's footprint and serve it over the ISP's internal network. - **Google edge points of presence:** Google-operated sites at peering locations and internet exchanges. They serve viewers whose ISP has no GGC node, and they also fill GGC nodes on a miss. - **Regional clusters:** Larger caches in Google's own network that hold the broader catalog, rarely requested renditions and freshly uploaded videos. - **Origin storage:** Durable storage of the uploaded masters and every encoded rendition. It almost never serves viewers directly. Fill is partly reactive (a miss pulls the object down the hierarchy) and partly proactive. Predictable demand, such as a scheduled premiere or a video trending in a region, can be pushed toward the edge before the peak, turning what would be a peak-hour upstream transfer into a local cache hit. ## **Google Global Cache: what ISPs actually host** Google Global Cache is a program in which Google supplies cache servers to qualifying ISPs, who provide rack space, power and network connectivity. The nodes serve YouTube and other cacheable Google content to that ISP's subscribers. Google operates and fills the servers; the ISP's benefit is reduced upstream traffic and lower interconnect cost. The arrangement shifts the economics for both sides. The ISP stops paying, in transit or in peering port capacity, for the same popular video crossing its border thousands of times an hour. Google stops paying for long-haul delivery and gets a cache one network hop closer to the viewer, which shortens the path that TCP or QUIC congestion control must probe. ### Worked example: what a GGC node saves an ISP The arithmetic below is this blog's estimate with stated assumptions, not a published figure. Assume an ISP whose YouTube traffic peaks at 200 Gb/s, a GGC hit ratio of 85% and an upstream cost of $0.20 per Mbps per month at the 95th percentile. - Without GGC: 200 Gb/s crosses the border, which is 200,000 Mbps ร— $0.20 = $40,000 a month. - With GGC at 85% hit ratio: only the 15% miss traffic (30 Gb/s) crosses the border, which is 30,000 Mbps ร— $0.20 = $6,000 a month. - Avoided cost: about $34,000 a month, plus 170 Gb/s of headroom freed on upstream ports during the evening peak. The hit ratio is the sensitive input. At 70% the saving falls to $28,000 a month; at 95% it rises to $38,000. Every point of hit ratio moves about $400 a month at this scale, which is why rendition sprawl (covered below) is a real cost, not an aesthetic concern. ## **How YouTube delivers video: the request flow** When a viewer presses play, the player first fetches video metadata and a list of available streams from YouTube's application servers. Those stream URLs point to googlevideo.com hostnames that encode a specific cache node or node group. The node choice is made by Google's mapping system before the first media byte is requested. The mapping uses the client's IP address and network, node health, current load and whether a GGC node exists inside the viewer's ISP. If the selected node misses, it fills from a Google edge or regional cluster rather than redirecting the client, so the viewer keeps one connection to one host. For on-demand playback, YouTube serves audio and video as separate streams and the player fetches them in byte ranges, adjusting rendition as throughput and buffer level change. That is adaptive bitrate (ABR) streaming in the DASH style rather than classic fixed-segment HLS. Live streams use segmented delivery, where segment duration directly trades latency against request rate. ### QUIC and HTTP/3 on the delivery path YouTube was one of the earliest large deployments of QUIC, now standardized as HTTP/3 over QUIC (RFC 9000 and RFC 9114). Google's 2017 published measurements reported that QUIC reduced YouTube rebuffer rates by roughly 15% on mobile and 18% on desktop compared with TCP and TLS. The gains come from faster connection setup, connection migration across network changes and no transport-level head-of-line blocking under packet loss. ### Segment and chunk sizing For live and low-latency streams, two-second segments split into sub-second chunks shorten startup and give the ABR algorithm faster bandwidth feedback. On-demand libraries usually prefer four to six seconds, which halves or thirds the request count per viewer-minute and improves encoder efficiency. These are general engineering ranges, not YouTube-published values. ## **YouTube video formats: H.264, VP9 and AV1** YouTube encodes each upload into a ladder of resolutions and bitrates across several codecs: H.264 for universal compatibility, VP9 (deployed at scale since the mid-2010s) and AV1 for capable devices, with Opus and AAC for audio. The player picks the most efficient codec the device can hardware-decode. Encoding effort follows demand. A video watched a few dozen times does not justify the compute cost of a full AV1 ladder; a video watched tens of millions of times does, because the bitrate saving multiplies across every view. As typical ranges, VP9 saves 30 to 50% bitrate against well-tuned H.264 at similar quality, and AV1 saves a further 20 to 30% against VP9. Google disclosed in 2021 that its custom video transcoding chip improved encoding efficiency by 20 to 33 times over its previous CPU-based fleet, which is what makes AV1 economically viable at YouTube's scale. The trade-off sits on the client. Software AV1 decoding drains batteries and drops frames on older phones, so a correct codec decision depends on device decode capability, not just bandwidth. VVC (H.266) remains niche in 2026 because hardware decode support is still thin. ## **YouTube CDN vs. Netflix Open Connect** Both companies embed caches inside ISPs, but their catalogs force different designs. Netflix runs a finite, curated catalog and fills its Open Connect Appliances proactively during nightly off-peak windows based on predicted popularity. YouTube serves an unbounded user-generated catalog where most videos are long tail, so it relies more on reactive fill and a deeper hierarchy. | Dimension | YouTube (Google Global Cache) | Netflix Open Connect | | --- | --- | --- | | Catalog shape | Unbounded, user-generated, heavy long tail | Finite, licensed and original titles | | Fill strategy | Mostly reactive on miss, with prefill for predictable demand | Mostly proactive, in off-peak fill windows | | Content scope | YouTube plus other cacheable Google content | Netflix video only | | Fallback tier | Google edge PoPs and regional clusters on Google's backbone | Open Connect appliances at internet exchanges | | Live content | Core workload, large daily volume | Occasional events on a VOD-first design | The catalog decides the design: a predictable catalog rewards proactive fill, while a long-tail catalog needs a deep hierarchy that absorbs misses cheaply. Most streaming platforms sit closer to Netflix than to YouTube. If your top 5% of titles carry most of your views, scheduled prefill of those titles beats a deeper cache tree. ## **What breaks in YouTube-style delivery, and how to confirm it** Each failure below shows up as a viewer symptom before it shows up on a server dashboard. The confirmation step is what separates a CDN problem from a client or network problem. - **Cold cache at event start:** a premiere or match begins and thousands of viewers miss on a node at once. Symptom: time to first frame spikes in one ISP or region. Confirm: origin or mid-tier request rate climbs in step with edge misses, and request coalescing counters show collapsed fills. - **Congested ISP aggregation:** the cache is inside the ISP but the link behind it is saturated at peak. Symptom: rebuffering by time of day, concentrated in one ASN, while cache hit ratio stays high. Confirm: per-ASN throughput drops at the same hour every evening. - **Mapping drift:** a routing or resolver change sends viewers to a distant node. Symptom: higher startup latency with no error increase. Confirm: the serving node or PoP identifier in response headers changes for a stable client population. - **Rendition sprawl:** six resolutions across three codecs is 18 video renditions per title, and each one dilutes cache capacity. Symptom: hit ratio declines as the codec mix grows. Confirm: hit ratio broken down by codec and resolution, not just in aggregate. - **Decode failure on the client:** the server delivers AV1 to a device that falls back to software decoding. Symptom: dropped frames and playback failures with healthy CDN metrics. Confirm: hardware decode success rate segmented by device model and codec. ### Verify this on your own stack Open YouTube's Stats for nerds overlay during playback: it shows the active codec (av01, vp09 or avc1), connection speed and buffer health. In browser developer tools, the media requests reveal the googlevideo.com host serving them, which identifies whether you hit a node inside your ISP. On your own platform, track time to first frame, rebuffer ratio and hit ratio per ASN and per codec. ## **What smaller video platforms should copy** Almost no platform can place hardware inside ISPs, but most of YouTube's gains come from decisions that do not require it. | YouTube technique | Copy it when | Skip it when | | --- | --- | --- | | Demand-based AV1 encoding | A small set of titles carries most views | Views are flat across the catalog and encode cost dominates | | Scheduled cache prefill | Releases and events are predictable | Demand is unpredictable, user-generated content | | Origin shield and request coalescing | Always, for any multi-PoP delivery | Never worth skipping | | Codec ladder trimming | Hit ratio falls as renditions grow | The audience is on one device class | Origin shielding and coalescing pay off for every platform, while AV1 and prefill pay off only when demand is concentrated and predictable. Prefill only the most-watched renditions and the opening minutes of predictable events; warming a full ladder wastes cache on renditions few devices will request. BlazingCDN's [**Video CDN for on-demand libraries**](https://blazingcdn.com/video-cdn/) replicates the library inside the CDN and keeps the origin out of the delivery path, with origin shield, request coalescing and cache warm-up through the API included; it serves over HTTP/2 and TLS 1.3 and does not support HTTP/3. ## **FAQ: YouTube CDN and Google Global Cache** ### Is the CDN YouTube uses its own network or a third party? YouTube uses Google's own delivery network, not a third-party CDN. Traffic is served from Google Global Cache nodes inside ISPs, Google edge points of presence and regional clusters on Google's backbone. Owning every tier lets Google tune mapping, cache fill, transport protocols and codec selection together, something a platform renting a commercial CDN controls only partially. ### What is Google Global Cache? Google Global Cache is a set of Google-owned cache servers installed inside ISP networks that serve YouTube and other cacheable Google content to that ISP's subscribers. The ISP provides space, power and connectivity, and Google operates and fills the servers. Popular videos are served locally, which cuts the ISP's upstream traffic and shortens the delivery path for viewers. ### How does YouTube deliver video to a viewer? YouTube delivers video by mapping each viewer to a specific cache node, then streaming separate audio and video tracks in byte ranges with adaptive bitrate selection. The mapping considers the viewer's network, node load and whether a Google Global Cache node exists in their ISP. Misses fill from Google's edge or regional clusters without redirecting the player. ### Does the YouTube CDN use AV1 or VP9? The YouTube CDN delivers both, plus H.264 for compatibility. The player chooses the most efficient codec the device can decode in hardware. Popular videos are more likely to receive AV1 encodes because the bitrate saving multiplies across many views, while less-watched videos may only have VP9 and H.264 renditions available. ### How is Google Global Cache different from Netflix Open Connect? Both place caches inside ISPs, but Google Global Cache handles an unbounded user-generated catalog and relies more on reactive fill through a deep hierarchy. Netflix Open Connect serves a finite catalog and fills appliances proactively during off-peak windows based on predicted popularity. Google Global Cache also serves other Google content, while Open Connect serves only Netflix. ## **Measure your own cache offload, then test delivery on real traffic** This week, pull a month of delivery logs and compute hit ratio per ASN and per codec, alongside time to first frame and rebuffer ratio for the same segments. Then rerun the offload arithmetic above with your peak bandwidth and your real hit ratio: it shows what each point of hit ratio is worth to you and whether trimming the rendition ladder or prefilling top titles pays off first. If your library is on-demand video and delivery cost is the next lever, the [**BlazingCDN OTT and VOD streaming CDN**](https://blazingcdn.com/ott-vod-streaming-cdn/) page describes a 14-day testing period on real production traffic at $5 per TB with the monthly minimum waived. Share: [f](https://www.facebook.com/sharer/sharer.php?u=https://blog.blazingcdn.com/en-us/how-youtube-cdn-works) [in](https://www.linkedin.com/sharing/share-offsite/?url=https://blog.blazingcdn.com/en-us/how-youtube-cdn-works) [๐•](https://twitter.com/intent/tweet?url=https://blog.blazingcdn.com/en-us/how-youtube-cdn-works&text=) [โœ‰](mailto:?subject=%3Cspan%20id="hs_cos_wrapper_name"%20class="hs_cos_wrapper%20hs_cos_wrapper_meta_field%20hs_cos_wrapper_type_text"%20style=""%20data-hs-cos-general-type="meta_field"%20data-hs-cos-type="text"%20%3EHow%20YouTube's%20CDN%20Works:%20Google%20Global%20Cache%20and%20Video%20Delivery%3C/span%3E&body=https://blog.blazingcdn.com/en-us/how-youtube-cdn-works) ![BlazingCDN](https://blog.blazingcdn.com/hs-fs/hubfs/Logo/blog-logo-w.png?height=24&name=blog-logo-w.png) *Heavy traffic.* Light bill. The CDN for video and large traffic Their monthly bill vs ours - 20 TBFastly $2,087**$92.50** - 50 TBCDN77 $990**$215** - 200 TBCloudFront $11,965**$765** Published list prices, Aug 2026 [Calculate your cost](https://blazingcdn.com/cdn-cost-calculator/?utm_source=blog&utm_medium=sidebar&utm_campaign=blog_sidebar&utm_content=compare_calc) ## Related posts [![](https://blog.blazingcdn.com/hubfs/Gemini-Blog/image-Nov-01-2026-03-56-37-9282-PM.jpeg)](https://blog.blazingcdn.com/en-us/multi-cdn-strategies-for-global-reach?hsLang=en-us) Learn ### [Global CDN: How to Deliver Content Worldwide With One or More CDNs](https://blog.blazingcdn.com/en-us/multi-cdn-strategies-for-global-reach?hsLang=en-us) A global CDN has to keep cached bytes within roughly 20 to 40 ms of round-trip time from users on every continent you ... Nov 1, 2026, 4:59:08โ€ฏPM [Read more](https://blog.blazingcdn.com/en-us/multi-cdn-strategies-for-global-reach?hsLang=en-us) [![](https://blog.blazingcdn.com/hubfs/Gemini-Blog/image-Nov-01-2026-03-25-10-9210-PM.jpeg)](https://blog.blazingcdn.com/en-us/what-is-multi-cdn-strategy-and-why-use-it?hsLang=en-us) Learn ### [What Is Multi-CDN? Strategy, Switching and When to Use It](https://blog.blazingcdn.com/en-us/what-is-multi-cdn-strategy-and-why-use-it?hsLang=en-us) Multi-CDN means serving the same hostnames and content through two or more CDN providers at once, with a steering layer ... Nov 1, 2026, 4:27:41โ€ฏPM [Read more](https://blog.blazingcdn.com/en-us/what-is-multi-cdn-strategy-and-why-use-it?hsLang=en-us) [![BlazingCDN](https://blog.blazingcdn.com/hubfs/Logo/blog-logo-w.png)](https://blog.blazingcdn.com?hsLang=en-us) [๐Ÿ“˜ Learn](https://blog.blazingcdn.com/cdn-learn?hsLang=en-us) [๐Ÿ“Š Benchmarks](https://blog.blazingcdn.com/cdn-ratings-and-benchmarks?hsLang=en-us) [๐ŸŽฌ Video & Streaming](https://blog.blazingcdn.com/video-streaming-cdn?hsLang=en-us) [๐Ÿญ Industries](https://blog.blazingcdn.com/cdn-industry-insights?hsLang=en-us) [๐Ÿ’ฐ Pricing & Costs](https://blog.blazingcdn.com/cdn-pricing-and-cdn-costs?hsLang=en-us) [โšก Compare](https://blog.blazingcdn.com/cdn-comparison?hsLang=en-us) [๐Ÿ”’ Security](https://blog.blazingcdn.com/cdn-security?hsLang=en-us) [๐Ÿ› ๏ธ Tools](https://blog.blazingcdn.com/cdn-tools?hsLang=en-us) [โœ๏ธ Publish with us](https://blog.blazingcdn.com/publish-with-us?hsLang=en-us) in f ๐• โœ‰ Copyright ยฉ BlazingCDN |. All rights reserved. ```json { "@context" : "https://schema.org", "@id" : "https://blazingcdn.com/#organization", "@type" : "Organization", "description" : "High-volume CDN for video, live streaming, OTT/IPTV, software, games, SaaS and large-file delivery.", "logo" : { "@id" : "https://blazingcdn.com/#logo", "@type" : "ImageObject", "height" : 560, "url" : "https://blazingcdn.com/wp-content/uploads/2024/08/logo-560-560.png", "width" : 560 }, "name" : "BlazingCDN", "sameAs" : [ "https://www.linkedin.com/company/blazingcdn", "https://x.com/BlazingCdn", "https://www.reddit.com/r/BlazingCDN/", "https://www.wikidata.org/wiki/Q141711074" ], "url" : "https://blazingcdn.com/" } ``` ```json { "@context" : "https://schema.org", "@id" : "https://blog.blazingcdn.com/#website", "@type" : "WebSite", "inLanguage" : "en-US", "name" : "BlazingCDN Blog", "publisher" : { "@id" : "https://blazingcdn.com/#organization" }, "url" : "https://blog.blazingcdn.com/" } ``` ```json { "@context" : "https://schema.org", "@id" : "https://blog.blazingcdn.com/en-us/how-youtube-cdn-works#article", "@type" : "BlogPosting", "author" : { "@id" : "https://blog.blazingcdn.com/en-us/author/blazingcdn#person", "@type" : "Person", "description" : "Vadym Borovskyi is Chief Product and Marketing Officer at BlazingCDN, where he has led product strategy and growth since 2021. He has more than 11 years of experience in IT product management and B2B marketing. On this blog he writes about CDN pricing and egress costs, video and software delivery, cache configuration, and how teams evaluate and switch CDN providers. He is based in Warsaw, Poland.", "image" : "https://143144902.fs1.hubspotusercontent-eu1.net/hubfs/143144902/vadym-borovskyi.jpg", "jobTitle" : "Chief Product and Marketing Officer", "knowsAbout" : [ "Content Delivery Networks (CDN)", "CDN pricing", "Video streaming delivery", "Software and game distribution", "Cache optimization", "Product Management", "B2B Marketing", "Digital Marketing", "Pay Per Click Marketing (PPC)", "Google Ads", "Search Engine Marketing (SEM)", "Marketing Strategy", "Artificial Intelligence (AI)", "Cloud Management", "SaaS Development", "IT Consulting", "Computer Networking", "Project Management", "Startups" ], "name" : "Vadym Borovskyi", "sameAs" : [ "https://www.linkedin.com/in/axelt/" ], "url" : "https://blog.blazingcdn.com/en-us/author/blazingcdn", "worksFor" : { "@id" : "https://blazingcdn.com/#organization" } }, "dateModified" : "2026-11-01T16:27:18Z", "datePublished" : "2026-11-01T16:27:18Z", "description" : "How YouTube delivers video: Google Global Cache inside ISPs, adaptive streaming, video formats and what makes YouTube's CDN so fast.", "headline" : "How YouTube's CDN Works: Google Global Cache and Video Delivery", "image" : "https://143144902.fs1.hubspotusercontent-eu1.net/hubfs/143144902/Gemini-Blog/image-Nov-01-2026-04-25-12-2357-PM.jpeg", "inLanguage" : "en-us", "isPartOf" : { "@id" : "https://blog.blazingcdn.com/#website" }, "mainEntityOfPage" : { "@id" : "https://blog.blazingcdn.com/en-us/how-youtube-cdn-works", "@type" : "WebPage" }, "publisher" : { "@id" : "https://blazingcdn.com/#organization" }, "wordCount" : 2332 } ``` ```json { "@context" : "https://schema.org", "@id" : "https://blog.blazingcdn.com/en-us/how-youtube-cdn-works#breadcrumb", "@type" : "BreadcrumbList", "itemListElement" : [ { "@type" : "ListItem", "item" : "https://blog.blazingcdn.com/en-us", "name" : "Blog", "position" : 1 }, { "@type" : "ListItem", "item" : "https://blog.blazingcdn.com/en-us/how-youtube-cdn-works", "name" : "How YouTube's CDN Works: Google Global Cache and Video Delivery", "position" : 2 } ] } ``` ```json { "@context" : "https://schema.org", "@id" : "https://blog.blazingcdn.com/en-us/how-youtube-cdn-works#faq", "@type" : "FAQPage", "mainEntity" : [ { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "YouTube uses Google's own delivery network, not a third-party CDN. Traffic is served from Google Global Cache nodes inside ISPs, Google edge points of presence and regional clusters on Google's backbone. Owning every tier lets Google tune mapping, cache fill, transport protocols and codec selection together, something a platform renting a commercial CDN controls only partially." }, "name" : "Is the CDN YouTube uses its own network or a third party?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Google Global Cache is a set of Google-owned cache servers installed inside ISP networks that serve YouTube and other cacheable Google content to that ISP's subscribers. The ISP provides space, power and connectivity, and Google operates and fills the servers. Popular videos are served locally, which cuts the ISP's upstream traffic and shortens the delivery path for viewers." }, "name" : "What is Google Global Cache?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "YouTube delivers video by mapping each viewer to a specific cache node, then streaming separate audio and video tracks in byte ranges with adaptive bitrate selection. The mapping considers the viewer's network, node load and whether a Google Global Cache node exists in their ISP. Misses fill from Google's edge or regional clusters without redirecting the player." }, "name" : "How does YouTube deliver video to a viewer?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "The YouTube CDN delivers both, plus H.264 for compatibility. The player chooses the most efficient codec the device can decode in hardware. Popular videos are more likely to receive AV1 encodes because the bitrate saving multiplies across many views, while less-watched videos may only have VP9 and H.264 renditions available." }, "name" : "Does the YouTube CDN use AV1 or VP9?" }, { "@type" : "Question", "acceptedAnswer" : { "@type" : "Answer", "text" : "Both place caches inside ISPs, but Google Global Cache handles an unbounded user-generated catalog and relies more on reactive fill through a deep hierarchy. Netflix Open Connect serves a finite catalog and fills appliances proactively during off-peak windows based on predicted popularity. Google Global Cache also serves other Google content, while Open Connect serves only Netflix." }, "name" : "How is Google Global Cache different from Netflix Open Connect?" } ] } ```