Nobody builds their own CDN just to deliver software updates and operating system downloads, even if it does make the process faster and more reliable. We can conclude therefore that as Apple’s own CDN ramps up it will be delivering large amounts of data over it and that can only mean video. Even if the primary motive for building a CDN is quality of experience rather than cost saving, it would make no sense either economically or competitively to invest in massive CDN capacity costing more than $100 million, capable of carrying multiple terabits per second just to improve the admittedly patchy performance of its software updates, which have prompted numerous complaints from customers.
Establishing its own CDN does fit with Apple’s longstanding strategy of owning its whole ecosystem as its cloud services become an increasingly important part of its offering. When the principal role of the internet for Apple was for delivering updates and iTunes downloads, it could profitably rely on third party CDN companies like Akamai and Level 3, as it has been. But video consumes far more bandwidth and when streamed is very demanding in terms of latency and jitter, which can only really be guaranteed with full control over the end to end infrastructure.
This does not necessarily mean owning the end to end path, which is impossible in the case of the last mile since that is in the gift of the local broadband service provider. But if you own the CDN you are much better placed to ensure delivery up to the interconnect with the local ISP and avoid that point becoming a bottleneck. Access bandwidth itself may or may not then be an issue, but at least all content providers will be in the same boat, providing net neutrality is upheld. In any case by owning the CDN it is possible to establish good service agreements with the ISP and ensure that the interconnects have plenty of capacity. Then as access network capacity continues to increase as fiber goes deeper and DSL vectoring improves, that will become a diminishing issue for delivery of OTT video, even at full HD, for more and more households across the developed nations at least.
As we reported a month ago, Alcatel-Lucent has shot an economic hole clean through fiber to the home technologies with its demonstration of 10 Gbps broadband speeds over the last 30 meters of twisted pair copper wiring. This is a major step forward because the last 30 meters is if anything more expensive to provision than all the rest of the last mile. It entails scheduling with customers, digging up gardens and drilling holes in walls, and is the most disruptive and expensive step in the creation of high speed broadband to the home in a Telco environment.
This is all why Netflix has been building its CDN for some years and has now established interconnection arrangements with ISPs like Comcast, AT&T and Verizon in the US. It is true that Netflix’s relationship with some of these ISPs has been factitious, but this partly reflects its business model, focusing heavily on price. For Netflix it is important that it pays as little as possible for interconnection to ISP access networks and preferably nothing at all, but for Apple that is not the case. Apple does not compete hard on price and is more relaxed about the interconnection agreements.