Apache Stratos uses two smart policies: Auto-scaling policy and Deployment policy. Stratos uses auto-scaling policy and deployment policy to calibrate the elasticity of the system. The auto-scaling and deployment policies are used to derive a cloud SLA from the platform as a service (PaaS) layer, which in turn benefits the software as a service (SaaS) layer applications. For example, if you want to derive five-9’s availability (99.999%), then you have to maintain a minimum instance count and run on different cloud vendors, different regions, different zones etc. The latter requirement can be fulfilled if defined in the deployment policies. 

The following sub-sections illustrate these two smart policies in detail.

Auto-scaling policy

An auto-scaling policy is a policy that determines the auto-scaling process, based on the load thresholds. The load thresholds in turn are determined based on the requests in flight, memory consumption and load average. The auto-scaler uses the latter mentioned parameters while taking decisions to scale up or down. For in-depth information on the JSON file that defines the auto-scaling policy, see the Auto-scaling Policy Resource Definition. 

Deployment policy

A deployment policy is a set of rules that will be applied when cartridges are deployed. The deployment policy describes as to which cloud partitions are used by a service. There can be multiple deployment policies that are compatible with a single application. Each deployment policy defines the applicable partitions, network partitions, partition algorithm, and the allowed maximum instances for each partition. For in-depth information on the JSON file that defines the deployment policy, see the Deployment Policy Resource Definition. 

  • No labels