Hot Route Management: Modify Live Streams Without Interruption

Add, remove, and modify outputs on live streaming routes with zero disruption. How Vajra Cast hot management works.

What is Hot Management?

Hot management is the ability to modify a live streaming route (adding outputs, removing outputs, changing destinations, enabling or disabling connections) without interrupting the stream or affecting other outputs on the same route. The stream keeps flowing to all unaffected destinations while you make changes.

This sounds like it should be a basic feature. It is not. Most streaming servers require a configuration reload, an application restart, or a complete route rebuild to change output destinations. During that restart, every output on the route drops for a few seconds. For a server handling 20 routes with 5 outputs each, a restart to change one output affects 100 streams.

Vajra Cast was designed from the ground up to support hot management on every route operation.

Why Hot Management Matters

Live Production Reality

During a live broadcast, requirements change constantly:

  • A sponsor requests their own dedicated feed 5 minutes before air
  • A social media platform needs to be added mid-show
  • An output is malfunctioning and needs to be removed
  • A CDN endpoint needs to be swapped for a different region
  • A test output that was used during rehearsal needs to be cleaned up
  • A backup recording destination needs to be added as insurance

Without hot management, each of these changes requires stopping the route, modifying the configuration, and restarting. During the restart, every output drops. In a live production, that is not acceptable.

With hot management, each change is atomic: it affects only the specific output being modified. All other outputs continue without interruption.

Multi-Tenant Environments

For platforms running multiple client streams through a single Vajra Cast instance (see OTT streaming), hot management is essential. Adding an output for Client A must not affect Client B’s streams. Without hot management, a configuration reload for one client would momentarily interrupt all clients.

Operations Supported

Add an Output

Add a new output to a running route. The existing stream is immediately distributed to the new destination alongside all current outputs.

Use cases:

  • Adding a social media platform during a live event
  • Adding a backup recording destination
  • Adding a monitoring tap for quality verification

The new output connects and begins receiving the stream within seconds. No other outputs are affected.

Remove an Output

Remove an output from a running route. The connection to that destination is closed cleanly, and all other outputs continue without interruption.

Use cases:

  • Removing a test output after rehearsal
  • Disconnecting a failed destination to clean up error logs
  • Removing a temporary feed after an event segment ends

Enable / Disable an Output

Toggle an output on or off without removing it from the route configuration. A disabled output maintains its settings but stops sending data. Re-enabling it resumes the stream.

Use cases:

  • Pre-configuring outputs before a broadcast and enabling them at showtime
  • Temporarily pausing an output while the destination is being reconfigured
  • Disabling an output during a commercial break and re-enabling for the next segment

Move an Output

Change the destination of an output. The old connection is closed and a new connection is established to the new destination. Other outputs are unaffected.

Use cases:

  • Swapping a CDN endpoint from one region to another
  • Redirecting a feed from a test server to the production server
  • Moving an RTMP output from one platform to another

Modify Output Settings

Change protocol-specific settings on an output (bitrate, encryption, latency) without removing and recreating it.

Use cases:

  • Adjusting SRT latency in response to changing network conditions
  • Updating an RTMP stream key for a platform that rotates keys
  • Changing HLS segment duration based on viewer feedback

How It Works Under the Hood

Vajra Cast’s architecture makes hot management possible through its internal routing design:

Zero-Copy Internal Multicast

When a stream is ingested, Vajra Cast creates an internal representation that multiple outputs can read from simultaneously. This internal multicast layer is the key to hot management:

Ingest --> Internal Buffer (shared memory)
              |
              |--> Output 1 (SRT): reads from buffer
              |--> Output 2 (RTMP): reads from buffer
              |--> Output 3 (HLS): reads from buffer
              |
              |--> [New Output 4]: connects to buffer (hot-added)

Adding a new output means creating a new reader on the existing buffer. Removing an output means closing a reader. Neither operation affects the buffer or the other readers.

Process Isolation

Each output runs as an independent process or thread. This means:

  • A failing output does not affect other outputs
  • Adding an output starts a new process
  • Removing an output terminates only that process
  • No shared state between outputs can cause cascading failures

API-Driven Operations

All hot management operations are available through the REST API:

# Add an output to route "abc123"
POST /api/routes/abc123/outputs
{
  "protocol": "rtmp",
  "url": "rtmp://live.twitch.tv/app/stream-key"
}

# Disable an output
PATCH /api/routes/abc123/outputs/output-5
{
  "enabled": false
}

# Remove an output
DELETE /api/routes/abc123/outputs/output-5

The same operations are available through the web dashboard with a visual interface.

Hot Management and Failover

Hot management also applies to input failover chains. During a live stream, you can:

  • Add a backup input to strengthen the failover chain
  • Remove a degraded input that is causing unnecessary switching
  • Reprioritize inputs by changing the failover order
  • Replace a failed input with a new source without rebuilding the route

This is particularly valuable for live events where network conditions change and backup sources need to be adjusted on the fly.

Practical Scenarios

Scenario 1: Adding a Platform Mid-Broadcast

Your live broadcast is streaming to YouTube and your website. The marketing team asks you to add Twitch and Facebook midway through the show.

Without hot management: Stop the route, add the outputs, restart. YouTube and website drop for 3-5 seconds. Viewers see buffering.

With hot management: Add the Twitch and Facebook outputs from the dashboard. YouTube and website streams are unaffected. Twitch and Facebook start receiving the stream within seconds.

Scenario 2: CDN Failover

Your primary CDN is experiencing latency issues. You need to switch to the backup CDN.

Without hot management: Stop the route, change the CDN endpoint, restart. All viewers disconnect during the switch.

With hot management: Add the backup CDN as a new output (it starts receiving immediately). Verify it is working. Disable or remove the primary CDN output. Total viewer impact: zero.

Scenario 3: Scaling During a Viral Event

Your stream is suddenly getting 10x the expected traffic. You need to add CDN endpoints in additional regions.

Without hot management: Each new CDN endpoint requires a restart, disrupting existing viewers.

With hot management: Add each new CDN endpoint as an output. Each connects independently without affecting existing outputs. Scale as needed, in real-time.

Scenario 4: Emergency Input Swap

During a live broadcast, your primary encoder develops issues. A technician has brought a replacement encoder online but it has a different IP address.

Without hot management: Modify the ingest configuration, restart the route. All outputs drop.

With hot management: Add the new encoder as a backup input. Once verified, promote it to primary. Remove the old input. The transition is smooth and controlled, and outputs remain uninterrupted throughout.

Monitoring Hot Changes

Every hot management operation is:

  • Logged with timestamps, operator identification, and the specific change made
  • Reflected in real-time on the dashboard (new outputs appear immediately, removed outputs disappear)
  • Exported as metrics to Prometheus (output count changes, connection events)
  • Available via API for integration with your operations tooling

This audit trail is critical for post-event review and troubleshooting.

Next Steps

← Back to main guide