May 2017 Summaries

YouTube has provided a platform for many filmmakers to share their work and build a fanbase. However, monetizing content effectively on YouTube remains challenging, especially for production teams with high costs. As a result, some content creators are evolving beyond YouTube by hosting their own content, building video platforms from scratch, developing subscription models, utilizing crowd-funding, and partnering with video streaming services. Successful content creators have honed skills in scriptwriting, voice acting, scheduling, directing, editing, and social media marketing. Some teams are already producing original content and developing business models to support themselves and their work. YouTube is where the cinematic talent of tomorrow is being nurtured, but they may not stay there forever as technology for hosting video and building platforms catches up.

A survey conducted in March 2017 with 1,035 US consumers aged between 18-44 revealed that the most frustrating problem experienced during video streaming on their TVs was rebuffering (47%). This was followed by video playback failure (19.9%) and slow video load times (18.8%). Respondents also reported that rebuffering was the most frequent issue (41.4%), while slow video start times were more common than playback failures or errors. Picture quality was considered the least frustrating and least frequent problem for users streaming video to their television sets. The majority of respondents (95.5%) strongly agreed that reliability and quality are very important to their video streaming experience through TVs, indicating a need for improvement in these areas.

The Flink Forward San Francisco 2017 conference highlighted many mission-critical applications powered by Apache Flink and revealed the direction of Flink platform development. Mux, a company that began using Flink in mid-2016, spoke at the event about their use of Flink for anomaly detection. The conference featured strong attendance from both attendees and sponsors such as Google, Alibaba, and Dell-EMC. Two notable sessions were Monal Daxini's keynote on Netflix's use of Flink for video view measurements and Stephan Ewen's talk on running Flink at a large scale. The event showcased the growing popularity of Flink among major organizations like Netflix, Uber, and Alibaba, as well as its potential for adoption by smaller businesses.

Failures in software and hardware systems are inevitable, regardless of the scale of operation. Most software outages occur during system updates or repairs, making failure detection and architecture crucial for minimizing downtime. Monitoring plays a significant role in identifying issues early on, with both active and reactive monitoring methods available. Active monitoring requires human oversight, while reactive monitoring relies on software to inspect the system state and alert humans when necessary. Alerting thresholds should be set appropriately for different metrics, and tiers can be used to separate urgent alerts from those that require less immediate attention. Postmortems are essential learning tools following an outage, helping teams understand what happened, how it could have been detected earlier, and whether the response was appropriate. Architecting against failure involves making informed decisions based on cost-benefit analyses, with strategies such as retry, backoff, rate limiting, caching, redundancy, buffering, reconsidering dependencies, and isolation being potential solutions to mitigate future failures. Continuous improvement in testing and release practices can also contribute significantly to reducing the likelihood of outages.

A survey conducted in March 2017 with 1,035 US consumers aged between 18-44 explored their experiences and feelings about video streaming services on their TVs. The most used streaming services were Netflix, YouTube, Amazon Prime, Hulu, and HBO. Consumers mostly blamed their Internet Service Provider (ISP) for slow load times and rebuffering, while they held the video app publisher responsible for poor picture quality and playback failures. Smart TVs were rarely blamed for streaming issues.