Resilience
and connectivity-loss behaviour.
"What happens to our rooms if the local AV network loses its uplink to the wide area network or the internet?"
01.The question, stated plainly
If the local AV network loses its uplink to the wide area network or the internet, while the departmental network remains healthy, can the rooms keep functioning? This post gives the direct answer, the architectural context, and the operating evidence behind it, plus the additional resilience patterns available if you want to harden the deployment further.
02.What actually happens
A presentation already running from a local laptop continues. Audio already routed continues. Displays already powered on stay on.
Touch panels freeze until connectivity returns. The room is frozen in its current state rather than falling into an unsafe one.
The control plane reconciles state automatically. No manual recovery step is required from your AV team.
03.The results speak for themselves
Control-plane availability sustained across 30+ enterprise tenants in higher education, healthcare, government, and corporate sectors.
In production at the University of Queensland, highly available AV control from the cloud for the past seven+ years.
04.How the architecture supports this
Four properties of the platform combine to deliver the behaviour described above. None of them are bolt-ons; all four are how Innomesh is built.
Redundancy in the connectivity layer
Your link to Innomesh runs over AWS Site-to-Site VPN or AWS Direct Connect, both of which carry built-in redundancy. Each Site-to-Site VPN connection terminates on two independent AWS endpoints in different Availability Zones, with both tunnels active, so traffic continues if one fails. Direct Connect deployments can be paired with a second Direct Connect circuit for redundancy, so the link itself is not a single point of failure.
Per-room container isolation across multiple data centres
Each room runs as its own containerised instance, deployed across multiple AWS Availability Zones (geographically separate data centres within the same AWS region). If one zone fails, the room's container is automatically routed to a healthy zone with no manual intervention. A single-room incident does not affect another room on the same campus.
Local media path, not through the Cloud
Innomesh sends control commands to your AV devices; it does not carry the audio or video stream itself. Audio and video flow directly between devices on your local network. So if the control plane becomes unreachable, what is already playing keeps playing.
Active-active control plane
The Innomesh control plane runs as an active-active deployment across multiple AWS Availability Zones. Zone-level failover is automatic. You don't need to provision failover infrastructure.
Local controller redundancy.
Clustered cloud control, plus one local failover. The local controller and AV devices in the room continue to operate independently of the cloud connection. Media stays on the local network, while control commands flow from the cloud when available, and from the local controller in a failover event.
05.Resilience options if you want to go further
Our consistent advice is to start with the standard pattern, observe behaviour through the pilot, and only adopt the additional patterns if real-world data points to them.
Dual Site-to-Site VPN tunnels across redundant firewalls
Suitable where you already operate redundant firewalls with HA failover. In production at one of our flagship Australian higher-education accounts.
Dual Direct Connect for HA
Two AWS Direct Connect circuits configured for high availability and dedicated to AV. In production at a flagship Australian higher-education account.
N+1 architecture at the control-plane edge
Customer-specific, and typically more expensive than the dual-tunnel pattern. Treated as an executive-comfort option rather than an operational requirement.
Innomesh inherits your connectivity resilience. It does not add a new dependency.
Microsoft Teams, OneDrive, Zoom, Panopto, lecture-capture scheduling, single-sign-on and identity federation already rely on the same wide-area connectivity that the AV control plane uses. Innomesh sits inside that envelope.
Black-box on-premise control systems hide their availability cost.
Controller reboots, firmware and physical interventions are real availability events, rarely tracked against an SLA. Cloud-native control replaces a fleet of independent failure surfaces with a single managed control plane held to an SLA.
06.Our availability promise
Response and resolution targets per severity, paired with a control-plane availability commitment measured over a rolling 12-month window. Excludes planned maintenance and third-party dependency outages.
Control-plane availability target
Measured over a rolling 12-month window. The target is contractual; the operating record across 30+ enterprise tenants supports it.
| Incident | Severity | First response | Resolution |
|---|---|---|---|
| Critical | 1 | 15minutes | 1hour |
| Severe | 2 | 30minutes | 8hours |
| Partial | 3 | 2hours | 48hours |
| General | 4 | 8hours | 72hours |
Walk through this with our team.
The Innomate team is happy to step through the connectivity pattern with your DevOps, Networks, Cloud and Cyber Security teams in detail. Reach out via your account manager, or contact us directly.