About the FGCP
FortiGate HA is implemented by configuring two or more FortiGate units to operate as an HA cluster. To the network, the HA cluster appears to function as a single FortiGate unit, processing network traffic and providing normal security services such as firewalling, security profile services, and VPN services.
HA cluster installed between an internal network and the Internet
Inside the cluster the individual FortiGate units are called cluster units. These cluster units share state and configuration information. If one cluster unit fails, the other units in the cluster automatically replace that unit, taking over the work that the failed unit was doing. After the failure, the cluster continues to process network traffic and provide normal FortiGate services with virtually no interruption.
Every FortiGate cluster contains one primary unit (also called the master unit) and one or more subordinate units (also called slave or backup units). The primary unit controls how the cluster operates. The role that the subordinate units play depends on the mode in which the cluster operates: (Active-Passive (AP) or Active-Active (AA).
The ability of an HA cluster to continue providing firewall services after a failure is called failover. FGCP failover means that your network does not have to rely on one FortiGate unit to continue functioning. You can install additional units and form an HA cluster.
A second HA feature, called load balancing, can be used to increase performance. A cluster of FortiGate units can increase overall network performance by sharing the load of processing network traffic and providing security services. The cluster appears to your network to be a single device, adding increased performance without changing your network configuration.
Virtual clustering extends HA features to provide failover protection and load balancing for Virtual Domains (VDOMs). See Virtual clusters.
FortiGate models that support redundant interfaces can be configured to support full mesh HA. Full mesh HA is a method of reducing the number of single points of failure on a network that includes an HA cluster. For details about full mesh HA, see Full mesh HA.
FGCP failover protection
The FGCP provides IP/MAC takeover for failover protection by assigning virtual MAC addresses to the primary cluster unit and then sending gratuitous ARP packets from the primary unit interfaces to reprogram the network.
Failover times can be less than a second under optimal conditions. You can fine tune failover performance for your network by adjusting cluster status checking timers, routing table update timers, and wait timers.
An HA cluster fails over if the primary unit fails (a device failure) or experiences a link failure. The cluster can detect link failures for connections to the primary unit using port monitoring and for connections between downstream network components using remote IP monitoring. To compensate for a link failover, the cluster maintains active links to keep traffic flowing between high-priority networks. Port and remote IP monitoring can be fine tuned without disrupting cluster operation.
FGCP session failover maintains TCP, SIP and IPsec VPN sessions after a failure. You can also configure session failover to maintain UDP and ICMP sessions. Session failover does not failover multicast, or SSL VPN sessions. Session failover may not be required for all networks because many TCP/IP, UDP, and ICMP protocols can resume sessions on their own. Supporting session failover adds extra overhead to cluster operations and can be disabled to improve cluster performance if it is not required.
Active-active HA load balances resource-intensive security profile features such as virus scanning, web filtering, intrusion protection, application control, email filtering and data leak prevention operations among all cluster units to provide better performance than a standalone FortiGate unit. If network traffic consists of mainly TCP sessions, the FGCP can also load balance all TCP sessions to improve TCP performance in some network configurations. On some FortiGate models you can also load balance UDP sessions. NP4 and NP6 offloading can accelerate HA load balancing (especially TCP and UDP load balancing). HA load balancing schedules can be adjusted to optimize performance for the traffic mix on your network. Weighted load balancing can be used to control the relative amount of sessions processed by each cluster unit.
Virtual clustering is an extension of the FGCP for a cluster of 2 FortiGate units operating with multiple VDOMS enabled. Not only does virtual clustering provide failover protection for a multiple VDOM configuration, but a virtual cluster can load balance traffic between the cluster units. Load balancing with virtual clustering is quite efficient and load balances all traffic. It is possible to fine tune virtual clustering load balancing in real time to actively optimize load sharing between the cluster units without affecting the smooth operation of the cluster.
Full Mesh HA
High availability improves the reliability of a network by replacing a single point of failure (a single FortiGate unit) with a cluster that can maintain network traffic if one of the cluster units fails. However, in a normal cluster configuration, single points of failure remain. Full mesh HA removes these single points of failure by allowing you to connect redundant switches to each cluster interface. Full mesh HA is achieved by configuring 802.3ad aggregate or redundant interfaces on the FortiGate unit and connecting redundant switches to these interfaces. Configuration is a relatively simple extension of the normal aggregate/redundant interface and HA configurations.
FortiOS HA provides a wide range of cluster management features:
- Automatic continuous configuration synchronization. You can get a cluster up and running almost as quickly as a standalone FortiGate unit by performing a few basic steps to configure HA settings and minimal network settings on each cluster unit. When the cluster is operating you can configure FortiGate features such as firewalling, content inspection, and VPN in the same way as for a standalone FortiGate unit. All configuration changes (even complex changes such as switching to multiple VDOM mode or from NAT/Route to Transparent mode) are synchronized among all cluster units.
- Firmware upgrades/downgrades. Upgrading or downgrading cluster firmware is similar to upgrading or downgrading standalone FortiGate firmware. The Firmware is uploaded once to the primary unit and the cluster automatically upgrades or downgrades all cluster units in one operation with minimal or no service interruption.
- Individual cluster unit management. In some cases you may want to manage individual cluster units. You can do so from cluster CLI by navigating to each cluster unit. You can also use the reserved management interface feature to give each cluster unit its own IP address and default route. You can use the reserved management interfaces and IP addresses to connect to the GUI and CLI of each cluster unit and configure an SNMP server to poll each cluster unit.
- Removing and adding cluster units. In one simple step any unit (even the primary unit) can be removed from a cluster and given a new IP address. The cluster keeps operating as it was; the transition happening without interrupting cluster operation. A new unit can also be added to an operating cluster without disrupting network traffic. All you have to do is connect the new unit and change its HA configuration to match the cluster’s. The cluster automatically finds and adds the unit and synchronizes its configuration with the cluster.
- Debug and diagnose commands. An extensive range of debug and diagnose commands can be used to report on HA operation and find and fix problems.
- Logging and reporting. All cluster units can be configured to record all log messages. These message can be stored on the individual cluster units or sent to a FortiAnalyzer unit. You can view all cluster unit log messages by logging into any cluster unit.
- FortiManager support. FortiManager understands FortiOS HA and automatically recognizes when you add a FortiOS cluster to the FortiManager configuration.
The FGCP uses a combination of incremental and periodic synchronization to make sure that the configuration of all cluster units is synchronized to that of the primary unit. This means that in most cases you only have to make a configuration change once to have it synchronized to all cluster units.
Some configuration settings are not synchronized to support some aspects of FortiGate operation. The following settings are not synchronized among cluster units:
- The FortiGate unit host name. Allows you to identify cluster units.
- HA override.
- HA device priority.
- Virtual cluster 1 and Virtual cluster 2 device priorities.
- The HA priority (
ha-priority) setting for a ping server or dead gateway detection configuration.
- The system interface settings of the FortiGate interface that becomes the HA reserved management interface.
- The default route for the reserved management interface, set using the
ha-mgmt-interface-gatewayoption of the
config system hacommand.
- The dynamic weighted load balancing thresholds and high and low watermarks.