You need to design a system that transcodes live video streams into multiple quality levels for adaptive bitrate streaming. Which architectural component is essential to handle the real-time transcoding workload efficiently?
Think about how to handle many streams at once without delay.
A distributed transcoding cluster allows parallel processing of video segments, enabling real-time handling of multiple live streams. A single server or client-side transcoding cannot scale well. A database is for storage, not processing.
You have a video library of 1000 videos, each 1 hour long at 1080p. You want to store 4 bitrate versions per video (240p, 480p, 720p, 1080p). If the average bitrate for 1080p is 5 Mbps, and lower bitrates scale proportionally (e.g., 720p is 3 Mbps), estimate the total storage needed in terabytes (TB). Which is the closest estimate?
Calculate total bits per video for all bitrates, convert to bytes, then multiply by number of videos.
Calculate each bitrate's data size for 1 hour, sum them, then multiply by 1000 videos. 5 + 3 + 1.5 + 0.75 Mbps = 10.25 Mbps total. 10.25 Mbps * 3600 seconds = 36,900 Mb = 4.6 GB per video. 4.6 GB * 1000 = 4.6 TB. Considering overhead and rounding, closest is 10 TB.
For video-on-demand (VOD) content, you can transcode videos either as a whole file continuously or in small chunks. What is the main tradeoff when choosing chunked transcoding over continuous transcoding?
Consider startup delay and storage management.
Chunked transcoding breaks video into small segments, enabling playback to start sooner and easier adaptive bitrate switching. However, it creates many small files, increasing storage overhead and management complexity. Continuous transcoding processes the whole file at once, delaying playback start but reducing file fragmentation.
In adaptive bitrate streaming, the client switches video quality based on network conditions. Which factor primarily influences the client's decision to switch to a lower bitrate stream?
Think about what affects playback smoothness.
The client monitors its buffer and download speed to decide if it should reduce bitrate to avoid playback stalls. Video length and display resolution do not directly affect bitrate switching. The number of available bitrates affects options but not the decision itself.
You observe frequent playback buffering and high latency in your adaptive bitrate streaming service during peak hours. Which system component is most likely the bottleneck causing these issues?
Consider where network delays and request handling happen during peak load.
During peak hours, CDN edge servers can become overloaded, causing delays in delivering video segments, leading to buffering and latency. Transcoding happens before streaming and is less likely to cause real-time playback issues. Encoding algorithm affects quality and size but not latency directly. Client software issues would affect individual users, not widespread peak hour problems.