ASTRA Redundancy Architecture

ASTRA Serial Changeover Unit

For maximum redundancy, ASTRA servers used for master control or news playout can be mirrored.  Mirrored servers run the exact applications with the same playlists. Both servers operate as if they were the main on-line server controlling all the devices in the broadcast chain.

Devices in the broadcast chain are generally controlled by RS-422 or RS-232 serial control protocols or simple GPI triggers.  To switch from one server to another, the serial or GPI cables to the devices need to switch control from the main server to the backup server without interruption.

The ASTRA Serial Changeover Unit can switch up to 48 serial lines and 15 GPIs ports within a frame.  In the vast majority of cases, there is no on-air disruption.

ASTRA Serial Changeover Unit.

Note: devices that are controlled by APIs are also included in a changover to a backup, with changes effective immediately.

Reasons for switchovers

Maintenance – If a server needs to go off-line for periodic maintenance or upgrades, the operator can initiate the switchover event that will be seamless with no on-air interruption

Hardware Failure – there is a “hearbeat” connection between the two servers.  If the heartbeat is missing for five  frames, the backup server commands the switchover unit to switch.  In the unlikely event a command was sent during these five frames, if no acknowledgement is received from the device, the command would be resent.

Failover Event

Below is an example of a typical mirrored system for playout.  Two mirrored ASTRA servers run the playlist synchronously but only one is physically connected to the devices in the broadcast chain.  The ASTRA Serial Changeover unit switches serial and GPI control within one frame.  A heartbeat connection monitors the health of the servers and initiates a switchover if the connection is lost for five frames.

   
Mirrored ASTRA servers with automatic failover

Database Synchronization

All playlist data is replicated in real-time. Changes in clips metadata are replicated in periodical batches every three minutes to avoid unnecessary database operations.