Supermicro Rack Scale Design
Supermicro RSD addresses the challenges faced by current Cloud and Telco datacenters to efficiently scale operations on new as well as existing hardware investments. Supermicro provides a complete rack solution built on its highly optimized server and storage building blocks in standard racks. Supermicro RSD maximizes resource utilization through disaggregating compute, network and storage resources distributed in a rack or across multiple racks (called a pod). All Supermicro RSD building blocks share power and cooling resources to provide best PUE metrics in large scale data center deployments.
Enabling such efficiencies in existing environments empowers data center operators to manage the new form of basic building blocks i.e., pools of storage, compute and network resources across racks the same way individual server and storage nodes are managed until now.
Supermicro RSD rack level management software is based on Intel’s RSD framework, which provides the scale and efficiency for cloud operators to perform operations such as pooling and provisioning from day 1, in addition to much needed telemetry and maintenance functions of the pod to manage allocated resources in a large scale data center environment.
Supermicro RSD management framework is built on DMTF’s Redfish APIs (Supermicro is a promoting member of Redfish API standards group) and provides RESTful APIs for cloud providers to orchestrate and provision the new Intel® Xeon® Scalable Processors based X11 generation and all existing X10 generation server and storage systems, as well as Supermicro networking products. Some of the salient features are,
- Deep discovery including location id of compute, storage and network building blocks in the rack
- Cooling and Power Management in the rack
- Asset Information, hardware monitoring and provisioning of the application-ready building block
Together, these features enable the following capabilities on hardware resources,
- Deploying various OpenStack distributions on composable hardware infrastructure using industry standard RESTful API and sharing pooled NVMe storage rapidly across resourcesmultiple application hosts
- Reconfiguring the pooled resources based on application workloads and node failures
- Powering off unused resources from the pool