High Availability (HA) is a deployment model that pairs redundant devices or nodes – commonly an active and a standby, sometimes an active-active pair – so that if one fails, the other continues delivering service with little or no interruption. Versa applies HA to branch appliances, Director, Analytics, and Controller nodes alike.
Key takeaways
- HA pairs redundant devices so a standby takes over if one fails.
- Two core models: active-standby failover and active-active load-sharing.
- VRRP supplies the virtual IP that hides a failover from clients.
- Versa applies HA to branch appliances, Director, Analytics, and Controllers.
- Failover speed, not zero downtime, is what HA actually guarantees.
On this page: Why it mattersHow it worksTypesHigh Availability & VersaFAQ
Why is High Availability important?
Every network has components that, left unprotected, become a single point of failure: the branch device carrying WAN and LAN traffic, the controller terminating overlay tunnels, the management node holding policy and configuration. If any one of these goes down without a backup, the outage is total until someone intervenes – a bad trade for edges and headends expected to run continuously.
HA matters because it converts that hard stop into a brief handoff. A standby device or node keeps synchronized state and takes over automatically, so users and applications see, at worst, a short interruption instead of a full outage. It also decouples maintenance from downtime: a software upgrade, certificate rotation, or hardware swap can happen on one node of a pair while the other keeps serving traffic.
What is High Availability?
An HA deployment pairs two (or more) devices or nodes that share configuration and, for stateful designs, live session state. Each side continuously monitors the other’s health – over a dedicated link, a routing adjacency, or a heartbeat protocol – and a defined trigger promotes the standby the moment the active side stops responding.
Clients should not need to know which physical node is active. That’s usually solved with a shared identity: a virtual IP address advertised by whichever node holds the active role, most often through the Virtual Router Redundancy Protocol (VRRP). Each candidate interface carries a priority (range 1–255, default 100); the highest-priority interface becomes master and owns the virtual address until it fails or is preempted. Behind that shared address, routes, sessions, and security associations are kept synchronized so the takeover doesn’t restart existing traffic.
Types of High Availability
Active-passive (active-standby) runs one node handling traffic while its pair stays idle, ready to promote itself on failure – simple, at the cost of standby capacity that sits unused until something breaks. Active-active runs both nodes simultaneously, each carrying traffic and backing up the other; it uses hardware more efficiently but demands tighter state synchronization. Clustering / N+1 redundancy extends the pattern beyond two nodes, so any surviving member can absorb a failed peer’s load – typical where headroom needs to exceed a single backup.
Recovery behavior varies too: non-revertive designs leave a recovered node as standby rather than reclaiming the active role automatically; revertive designs promote it back to active once healthy, after a configurable delay.
High Availability & Versa
Versa applies HA across the stack, not just to branch devices. For branch appliances, an HA pair requires two identical devices – same model, same software license tier – wired with a cross-connect port and a shared LAN port carrying VRRP; the VRRP virtual IP and priority determine which device is primary. VOS pairs also support stateful interchassis HA, synchronizing control-plane and data-plane state so failover doesn’t drop in-flight sessions, and raising quorum alarms if a standby loses contact with its active peer.
Director nodes run active-standby HA with data synchronized by Postgres replication. Administrators configure HA behavior – failover timeout, standby start timeout, revertive or non-revertive switchover – and can trigger a manual failover. The same pattern extends to Analytics aggregator nodes, Versa Messaging Service primary/secondary pairs, and cloud deployments of VOS on AWS, Azure, and OCI, all managed through Director, Concerto, or Titan Portal.
Frequently asked questions
What is the difference between active-standby and active-active HA?
Active-standby keeps one node idle until the active node fails, then promotes it. Active-active runs both nodes simultaneously, sharing traffic while backing each other up. Versa Director HA defaults to active-standby; Versa branch appliances also support active-active dual-CPE deployments over VRRP and cross-connect links.
What is VRRP and how does it relate to HA?
VRRP lets a device pair share one virtual IP address; the interface with the highest priority (1–255, default 100) becomes VRRP master. In Versa branch HA, that virtual IP backs the shared LAN gateway so clients see one continuous address across a failover.
Does HA eliminate all downtime?
No. HA removes single points of failure and shortens disruption to a failover window, but it can’t prevent every incident – a shared misconfiguration or both nodes failing together can still cause an outage. Failover speed, not zero downtime, is what HA actually delivers.
What is the difference between failover and revertive behavior?
Failover is the standby taking over when the active node fails. Revertive behavior governs what happens next: a revertive design promotes the recovered node back to active automatically after a timeout; Versa Director’s non-revertive default leaves it as standby instead.
Does HA require identical hardware?
For paired appliances, yes – Versa requires the same device model and license tier on both units in the pair. Software-based HA, such as Director or Analytics nodes running as VMs, is more flexible but still expects matching software versions and configuration.
More on High Availability
If you want to learn more, these are sources we trust.
| NIST CSRC glossary |
|---|
| NIST’s glossary of information-security and infrastructure terms, including availability as one of the core security objectives. |
Related terms
If you’re reading about HA, these are worth looking at next.
| Appliance | Virtual IP Address | Controller |
|---|---|---|
| The physical or virtual VOS device that HA pairs, deployed as a branch, hub, gateway, or Controller node. | The shared address, often advertised by VRRP, that lets clients reach whichever paired device is currently active. | A VOS device providing SD-WAN control-plane functions, typically deployed in HA pairs for continuous overlay availability. |