Service Clustering + Media Sync

Continuity + Keeps past media, but spreads corruption and gaps remain

Executive Summary

Clustering with media synchronization provides an active-passive approach, where only one recorder service is active at a time, while the standby service takes over in the event of failure. Because media databases are mirrored between nodes, previously recorded content remains accessible after failover—an improvement over basic recorder failover. However, failover still requires tens of seconds to minutes, producing blind spots in both live monitoring and recorded streams. Automatic mirroring also creates a cybersecurity weakness: if one node is encrypted or corrupted, the issue propagates to the standby. Like other passive designs, efficiency is limited to a one-to-one ratio, demanding a dedicated standby for each active recorder.

Failover

Service clustering turns on the standby service node when the primary node fails. The solution does not provide stream-level failover, where only failed primary streams are started on the standby.

Service Clustering during failover

Consistency

The media database is synchronized between the two stacks, improving consistency.

Media Sync. improves consistency

• Provides access to content recorded before the failover.

• But the non-zero failover time inevitably creates gaps in media availability. During this window, live monitoring is interrupted, and recorded streams are incomplete, resulting in blind spots in both real-time situational awareness and post-event investigations. These forensic gaps undermine the reliability of video evidence and can weaken compliance and liability defense when edge recording and restoration are unavailable.

Content Gap

Cybersecurity

Unfortunately, automatic media database mirroring means that the media content on both nodes gets encrypted if either node is compromised.

Efficiency

Efficiency is 1:1, meaning one standby node can support one primary node failure. Two standbys are required to support two simultaneous primary node failures, and so on.