Skip to main content

Capacity and cost management

Vendors manage capacity

The vendor is responsible for allocating resources and designing scaling mechanism to meet their SLA. The vendor should consider doing so in a cost-effective way. The cost estimate should be provided to the customer as part of the negotiation process. This is an unavoidable step for building on-prem because the customer will pay for the hardware.

Containers and virtual machines running in customer’s appliance mirror its resource consumption (e.g. CPU and memory usage) back to their corresponding facades in the vendor’s cloud. The vendor can use that information to scale up/down the number of facades - thereby scaling up/down the number of containers and virtual machines running in the customer’s appliance.

  • The vendor is in control of scaling instances. Each time the vendor deploys a facade, an instance is concurrently deployed within the Isolated VPC of the appliance.

  • The operational state of the appliance's internal instance will synchronize with the vendor facade. If vendor facades are set to automatically scale according to CPU usage, auto-scaling will kick in to provision additional, or decommission existing, facades. This will, in turn, provision or decommission capacity inside the customer's appliance.

Customers set hardware limits

To control costs, customers can use their cloud’s native quota/budgeting features to set limits on the amount of capacity the their appliance can provision. If a vendor's auto-scaling rules hit this limit, they will fail to provision more hardware inside the customer's appliance. Of course, this can affect software performance.