Smarter by Design.Simpler for Business
TR7Enterprise Application Delivery & Security Platform
What is high availability (HA); at which layers is it provided in TR7?
Short answer
High availability (HA) is the design that keeps a service running without interruption or with a short interruption when a component fails (NIST · high availability); the goal is to eliminate the component whose failure alone stops the whole service (single point of failure, SPOF). In TR7, HA is provided at several layers. At the device layer, two TR7 devices form a cluster with HA Clustering & Sync: the configuration is kept the same on the two devices, and a VIP moves to the peer device when the device on which it is Master goes out of service. At the server layer, the health check removes a Backend Service that does not respond from the distribution; the server marked as Backup Service comes into service when all of the main servers are unavailable (on HTTP and L7 TCP vServices; it does not apply on a Network vService). At the link layer, Bond or LACP keeps the traffic going over the remaining members when a port goes down; gateway reachability can be monitored with Gateway Monitoring and take part in the VIP failover decision. Redundancy across data centers is provided at the DNS layer with GTM.
Where It Is Used in TR7
| Layer | Where | Field / label | Details |
|---|---|---|---|
| Device | Settings → Device Management → HA Clustering & Sync | Clustering enabled, Cluster ID, Manual Sync Mode | Is TR7 itself a single point of failure; how do I set up Load Balancer redundantly? |
| Device | Network → Interfaces → Add IP | Ip Type = VIP, Master/slave | How is a VIP defined after the cluster is set up? |
| Device | Network → Interfaces → Edit Interface | VIP Communication, VIP Transition Method | How does a cluster work? |
| Server | Backend Service form; Load Balancer → Profiles → Health Checks | Health Check, Backup Service | Removing a server that does not respond from the distribution; backup server (on HTTP and L7 TCP vServices) |
| Link | Network → Interfaces → Add Interface | Type = Bond / LACP | What is the difference between Bond and LACP? |
| Gateway | Network → Interfaces → Add Route → Default Gateway | Gateway Monitoring | What is Gateway Monitoring; how does it affect VIP failover? |
| Data center | GTM | GTM | For redundancy across data centers, should I use Load Balancer or GTM? |
Layers
| Failure | Effect without HA | TR7 equivalent |
|---|---|---|
| TR7 device out of service | All vServices stop | Cluster; the VIP moves to the peer device (What is failover; when does it happen in TR7?) |
| Backend Service does not respond | Requests going to that server get an error | The health check removes the server from the distribution; if needed, the Backup service comes into service (on HTTP and L7 TCP vServices) |
| Cable or switch port failure | All traffic of the interface is cut | Bond / LACP members; in a cluster, VIP failover with TR7 Link Check |
| Gateway unreachable | Outbound traffic from the route table is cut | VIP failover with Gateway monitoring + TR7 GW Check |
| Data center unreachable | All services in that center stop | GTM directs the client to a healthy data center |
Common Misconceptions
- "Setting up two TR7 devices is enough for HA." For the vService to be able to fail over, its Frontend IP must be a VIP, and there must also be redundancy at the server and link layers; otherwise another single point of failure remains.
- "Any number of devices can be added to a cluster." A TR7 cluster consists of two devices; Cluster ID can be only 1 or 2.
- "There is no interruption at all with HA." VIP failover happens with a short interruption; the failover time is determined by VIP Transition Duration (s) and the selected method.
How it is done in the documentation
