Nova Autoscalers change the way you deploy ADCs using Nova from a server to a service, dynamically scaling-in and scaling-out as your load requires.
Typically with Nova ADC you attach your ADC configuration to one or more Nodes manually, selecting which servers you want your ADC to run on. With Autoscalers you attach the ADC to an Autoscaler which dynamically shifts the number of servers based on load.
For example, you may say run at least 2 servers, and no more than 10 servers. In between that Nova ADC will launch and shutdown servers as it sees fit in order to handle your incoming traffic. During peaks it may decide it needs 8 servers and when it quietens down it will shut them down, saving you money, and handling bursts of traffic.
Autoscaling requires access to your Cloud in order to automatically launch new Nova Nodes, and to destroy them. It will make decisions on shutting down servers based on the last 10 minutes of information, and scaling-up servers based on the last 120 seconds of information.
You can decide the minimum and maximum, as well as the size of the instance to launch to control your costs.
Autoscalers automatically generate and maintain a Nova Destination for you, meaning that you are given a FQDN to send traffic to that will automatically scale. This highly available domain name will ensure that servers are added once ready, and removed once drained.
You attach your ADC config the same way as you normally would, but instead of to Nodes or specific Tags you can attach it to your new Autoscaler Tag, labelled "AS".
When Nova Autoscalers decide to terminate cloud instances it takes 5 minutes from the decision to the shutdown. The reason for this is to drain the servers of any active connections. The hosts will be removed from the dynamic destination FQDN immediately, and then shut down 300 seconds after that.