HP (Hewlett-Packard) 445946-001 ユーザーズマニュアル

 

 

 

High availability

 

 

 

  

 

175

 

A backup router can stop receiving advertisements for one of two reasons—the master can be down, or 

all communications links between the master and the backup can be down. If the master has failed, it is 

clearly desirable for the backup (or one of the backups, if there is more than one) to become the master. 

 

 

 If the master is healthy but communication between the master and the backup has failed, 

there will then be two masters within the virtual router. To prevent this from happening, configure 

redundant links to be used between the switches that form a virtual router.  

Failover methods 

With service availability becoming a major concern on the Internet, service providers are increasingly 

deploying Internet traffic control devices, such as application switches, in redundant configurations. 

Traditionally, these configurations have been hot-standby configurations, where one switch is active and 

the other is in a standby mode. A non-VRRP hot-standby configuration is shown in the figure below:  

 Non-VRRP hot-standby configuration  

 

While hot-standby configurations increase site availability by removing single points-of-failure, service 

providers increasingly view them as an inefficient use of network resources because one functional 

application switch sits by idly until a failure calls it into action. Service providers now demand that vendor 

equipment support redundant configurations where all devices can process traffic when they are healthy, 

increasing site throughput and decreasing user response times when no device has failed. 
The HP 10GbE switch high availability configurations are based on VRRP. The switch software 

implementation of VRRP includes proprietary extensions.  
The switch software implementation of VRRP supports the Active-Active mode of high availability.  

Active-Active redundancy 

In an active-active configuration, shown in the following figure, two switches provide redundancy for each 

other, with both active at the same time. Each switch processes traffic on a different subnet. When a 

failure occurs, the remaining switch can process traffic on all subnets.  
The following figure shows an Active-Active configuration example.