Historically, database environments would require configuration tuning to achieve predictable, consistent performance. However, cloud-native architectures such as Kubernetes discourage this approach from an application standpoint. This is because manual tuning can become unmanageable at large scale. The idea is that components should be as simple as possible and used in a commodity fashion. The Kubernetes design principle “cattle, not pets” is used to describe this. The PowerEdge, PowerStore, and Arc-enabled SQL MI stack has exceptional performance with no tuning required.
PowerEdge servers enable this with System Profiles, allowing you to select the best settings by choosing a profile for performance, power savings, and others, without the need to tune individual settings.
PowerStore storage appliances utilize machine learning to tune storage resources based on appliance usage, either on a single appliance or across multiple appliances in a cluster. This allows for optimal performance without manual configuration for individual storage volumes and hosts. For mixed workload environments, PowerStore performance policies can be used to set relative performance priority of high, medium, or low that take effect only if resources become constrained.
Dell Container Storage Interface (CSI) drivers for PowerStore allow storage to be provisioned dynamically and allow the storage to follow container workloads as they move between nodes in the cluster for load balancing or availability. Therefore, all common storage management tasks are performed by the Kubernetes control plane and Arc-enabled SQL MI configuration.
Arc-enabled SQL MI adapts to the workload by scaling within the deployment limits and utilizing the proven SQL Server core database engine. Compute and memory are managed with requests and limits upon deployment, and Arc-enabled SQL MI will scale within those boundaries to accommodate varying workloads.