两种权宜之计变成了长期问题

Our industry is prone to taking a quick-fix approach to solving short-term issues, 不经意地(或辉煌地), depending on whom you talk to) locking the technology into a “just good enough” solution for years or even decades. 通常, 解决方案是事实上的解决方案, meaning that one company or a small handful owns a key patent (or group of patents) to a particular technology that is neither an international standard nor backwards-compatible to international standards. 这种做法阻碍了该行业的长期创新.

在过去的两年里, I’ve had a chance to work with some very innovative compression technologies, which appear to be hampered by the industry’s reliance on two long-in-the-tooth de facto standards: HTTP segment streaming and buffer-based playback.

We all know that the basic premise for live streaming via HTTP is, at best, a flawed effort. However, the idea gained traction when two key trends coincided. 第一个, corporate customers complained that their IT departments were blocking streaming ports on routers and data switches, eliminating the ability to watch external live-streamed content inside the corporate firewall. 第二个, premium content providers were having difficulty scaling live-stream delivery since most real-time media protocols required expensive servers that typically only served fewer than 100 simultaneous viewers.

通过HTTP流解决了这些问题. The devil in the details, though, is the nagging permanence of longer-than-broadcast latency. Even as the average internet user’s data rates have increased, often to the point at which segmented HTTP manifest files don’t need to switch between renditions, we’re still holding on to the HTTP-segment crutch as we limp toward the finish line of global live-stream delivery.

第二个变通方法, 缓冲, is an even bigger impediment to streaming innovation and global synchronous delivery.

Unless you’ve been around the industry for the full 21-plus years of innovative chaos, 你可能从未听说过破裂.com and the David versus Goliath fight it waged successfully against Microsoft around the patented concept of buffer-based playback. 简而言之就是破裂.Com创建了一个实时视频传输的解决方案, 引起了罗伯特·X等权威人士的兴趣. 早在1994年，投资者就喜欢U2乐队. 1997年，U2乐队用这种技术播放了一场音乐会, it became apparent that acceptable visual quality with then-state-of-the-art codecs far exceeded the available bandwidth the average consumer had access to.

破裂.com创建了破裂ware, which monitored the bandwidth and watched for the frequency of stalls in a consumer’s video stream. 如果播放过于频繁地停止, 破裂ware would pause playback and inform the consumer that it was going to buffer enough of the stream to allow him or her to watch the remainder of the video uninterrupted—if only the consumer would be patient for several dozen seconds.

这解决了短期问题, but ultimately was used by consumers as a way to watch streams at data rates far above their available bandwidth, typically double the data rate of their dial-up modem or early DSL or T-1 internet connection.

Microsoft saw a demonstration of the technology and added it to Windows Media Player 9. 破裂之后成功起诉专利侵权. The idea of the streaming buffer had firmly caught hold at that point, leading Microsoft to buy a non-exclusive license for 破裂ware, thereby guaranteeing we’d be stuck with the concept of buffering for at least the next 2 decades of streaming.

总之, I want to challenge the industry to strongly consider jettisoning both of these workaround technology solutions. Without a bold move to eliminate HTTP-segment-based streaming and buffer-based playback, there’s little incentive to challenge the industry to innovate with 21st-century technologies. Rather than relying on de facto standards from the 20th century, it’s time for the industry to develop innovative solutions that solve the joint issues of zero-frame latency and global scalability.

[本文发表于2019年7月/ 8月号。 流媒体杂志 “太慢了”."]