October 2019 Summaries

The article discusses five strategies for coping with DNS failures and ensuring service reliability. It highlights that while a DNS outage may seem unlikely, recent statistics show that many companies are vulnerable due to using only one provider or having DNS servers on an internal network. The first strategy is understanding root server issues, as there are 13 root name servers responsible for delegating every domain and knowing what could go wrong helps in decision-making. The second strategy involves knowing the issues with TLD (top-level domain) authoritative servers, as not all domains have robust infrastructure to defend against attacks. The third strategy focuses on understanding name server issues and making smart DNS provider choices. The fourth strategy emphasizes taking a preventive approach to avoid renewal issues by regularly checking for upcoming domain expiration or SSL certificates going out of date. Lastly, the fifth strategy suggests safeguarding your ecosystem by using Anycast for everything and not coupling endpoints and DNS zone control in one provider.

WebRTC is a standard for real-time communication embedded in web browsers, enabling applications like voice or video calling and data sharing. However, it lacks signaling capabilities, which developers need to implement themselves. This involves creating a protocol for finding the person you want to communicate with and negotiating terms of communication. Developers can use various methods such as XMPP, MQTT, SIP, WebSocket, or Comet-type solutions like XHR. Alternatively, they can opt for third-party managed services like Ably Realtime, which takes care of uptime, monitoring, security, and scaling. There are three main approaches to developing communication with WebRTC: DIY, semi-managed, and fully managed. The semi-managed approach offers less vendor lock-in and more flexibility in mixing and matching components.

A technical review of Google's live score feature in search results reveals surprising use of old-school tech from the 90s, contrary to Google's reputation for efficient micro-optimization and pioneering initiatives like AMP. The reviewer found that Google uses HTTP polling instead of more suitable protocols such as Websockets or Server-Sent Events (SSE), resulting in an 80x increase in overhead compared to a raw Websocket solution. This approach also leads to high variable latencies and diminished user experience, with average latency for score updates being 25 times slower than any streaming transport that could have been used. Furthermore, Google's scoring synchronization is inefficient, sending the entire state object every 10 seconds instead of only the changes. The reviewer suggests using Xdelta as a delta encoding algorithm to improve efficiency and reduce bandwidth consumption.