𝗛𝗼𝘄 𝗜 𝗦𝗽𝗹𝗶𝘁 𝗠𝘆 𝗟𝗶𝘃𝗲𝘀𝘁𝗿𝗲𝗮𝗺 𝗔𝗿𝗰𝗵𝗶𝘃𝗲

𝗛𝗼𝘄 𝗜 𝗦𝗽𝗹𝗶𝘁 𝗠𝘆 𝗟𝗶𝘃𝗲𝘀𝘁𝗿𝗲𝗮𝗺 𝗔𝗿𝗰𝗵𝗶𝘃𝗲

I build tools at Shiftbloom Studio. I fixed a cost problem with media systems. Many systems use one big recorder. This costs money even when no one streams.

Live capture needs speed. Other tasks like VOD downloads or repairs do not. These tasks wait in a queue. They run on cheaper hardware.

I split the system into two parts:

• Observer cells: These handle live streams.
• Harvest cells: These handle the queue.

A small cron job manages the cells. It checks the queue size. It starts or stops cells based on need.

Observer cells record one channel. They write to object storage. They use a standby window. This stops broken VOD fragments.

Harvest cells do the heavy lifting. They download VODs and re-encode files. They run on any Docker host.

I treated live recording and backlog work as one problem before. They are different. One needs a specific assignment. One needs total throughput.

I moved 15.5 TB of data in 36 hours. Live streams did not drop a frame.

Separate urgent work from later work. Your system becomes simpler. It uses only what it needs.

Source: https://dev.to/fabianzimber/how-i-split-my-livestream-archive-at-shiftbloom-studio-48d3 Optional learning community: https://t.me/GyaanSetuAi