The very simple answer is that using two load balancers in a highly available (HA) configuration provides protection against downtime caused by outages. While the cost impact of an outage will vary by sector and organization size, the impact of an outage greatly outweighs the additional cost of deploying a HA load balancing environment.
The impact of downtime will vary across business. However, the following table gives an indication of the cost of an hour of application downtime.
|Organization Size||Hourly Downtime Cost|
The downtime cost numbers above are based on research from the Aberdeen Group (Building a fast lane to better data center performance - 2016).
It totally depends on your budget. If you are able to deploy two load balancers in HA mode, you already have an advantage in that you can take either load balancer down for maintenance without disruption to your website visitors. If your budget allows you to have two load balancers supported by a third development unit, it would be ideal, as you can test configurations before applying them to the production units. Some companies even deploy up to five load balancers - two for production use, two for testing new configurations ready to be deployed, and one spare for development work.
Compared to the cost of an outage, a HA load balancer environment makes financial sense.
Absolutely! Stateful failover means that the secondary backup load balancer keeps records of the sessions created by the primary load balancer. Should the backup unit need to take over from a failing primary unit, stateful failover allows the backup unit to continue to support the active sessions without service outage for the users.
Microsoft TechNet has many guides on how to configure High Availability for Microsoft applications and the Kemp Resource Library provides guidelines and templates for all the major workloads.