Active-Active Active-Passive configurations
Generally in cluster configurations and Load balancer Configurations, we come across the terms Active-Active and Active-Passive configurations. We will take a look at these configurations in more detail in this section.
Note that for simplicity, we have used two servers for explanation. The count is only for illustration purpose and could be higher.
Active-Active Configuration:
In this configuration, two servers would be online together and would independently share the workload and work as peers.
Fail over:
When one server goes down, the other server would cater to all the requests. In the above diagram, when Server B fails, server A would cater to the requests.
Active-Passive Configuration:
In this configuration, only one server would be online at any given time and the other server would be only configured as a fail over server and would remain offline (or passive). When the primary server (Server A, in the above diagram) goes down, the second server would come online and start catering to the requests.
Active-Active Vs Active-Passive configurations:
Active-Active | Active-Passive |
---|---|
# Of servers online in a happy day scenario < # of servers online in a fail over scenario. | # Of servers online in a happy day scenario = # of servers online in a fail over scenario. |
System performance suffers during fail over | System performance remains same during fail over |
Reference Implementations:
Load balancers could be configured to work either in Active-Active or in Active-Passive configurations. Similarly, few relational databases support either of these configurations. Active-Active or Active-Passive configuration could be used as a general design principle in System design.