ScalerNode Network
Hardware providers contribute physical servers with specific hardware classes to the ScalerNode Network.
Last updated
Hardware providers contribute physical servers with specific hardware classes to the ScalerNode Network.
Last updated
ICN defines Hardware Classes as optimized combinations of hardware resources that cater to specific use cases. By standardizing these classes, ICN ensures homogeneity across diverse hardware configurations provided by vendors. For instance, when offering ScalerNodes optimized for storage, ICN specifies tailored configurations for storage capacity, disk I/O operations and vCPU count. In contrast, accelerated computing ScalerNodes would prioritize GPU quantity and model, as well as memory allocation.
ScalerNodes with the same hardware class guarantee equivalent hardware resources, such as identical number of disks and storage capacity. This consistency is maintained even when devices from different manufacturers are used, although performance may vary slightly. If significant differences arise between ScalerNode, they are classified into separate classes.
During registration, Hardware Providers assign a specific hardware class to ScalerNodes. These nodes are then grouped according to their hardware class. Currently, only one storage class is available in ICN, simplifying initial deployment and focusing on storage capacity optimization.
ScalerNodes are registered within regions depending on their geographical location, ensuring data locality and compliance with regional regulations. Within a region, ScalerNodes are then assigned to clusters based on their hardware class and other economic factors.
Region: The highest level of geographic scope, representing a location or metropolitan area. When registering a new ScalerNode, Hardware Providers select one region from the available options in ICN. Within a Region, all ScalerNodes share certain economic parameters such as subsidy rewards or collateral requirements for each specific hardware class.
Cluster: A logical group within a region containing ScalerNodes from one or more Hardware Providers belonging to the same hardware class. Every cluster defines the maximum price that a ScalerNode can have for the specific hardware class. Builders book capacity in units of ScalerNodes at the cluster level.
Hardware Providers are responsible for the management and configuration of the physical servers and the physical network, ensuring secure remote access to the ScalerNodes. Hardware provider’s main set of responsibilities are:
Physical host maintenance including firmware updates and maintenance of attached devices such as storage drives, RAM, GPUs, etc.
Boot configuration via PXE or iPXE to enable automated bootstrapping of the host operating system.
Internal network configuration and management including routers, switches, and firewalls.
External network connectivity to the Internet and ensure reachability from outside.
Hardware Providers register ScalerNodes in ICN by specifying the following parameters:
Hardware Class: Type of hardware class* of the ScalerNode.
Location: Geographic location of the ScalerNode following ISO 3166.
Capacity: The total capacity of the ScalerNode for a certain hardware class*.
Rewards Share: Percentage of the rewards shared with delegators from 0 to 100%.
ReservationPrice: Price of the ScalerNode in ICNT per unit of capacity* per day.
Maximum booking duration: Maximum booking duration (by default 180 days).
*Currently, only one storage class is available with unit of capacity in terabytes (TB).
As part of the registration process, Hardware Providers are required to collateralize their ScalerNodes (please refer to the HP collateral requirements section for details).
Upon successful on-chain registration and collateralization, ScalerNodes are provisioned by granting access for Daemon deployment, enabling their integration into the ICN Protocol. The Daemon serves as the core diagnostic agent, collecting and reporting hardware-related performance metrics to the HyperNodes.
After successful on-chain registration, the following steps are carried out:
Capacity verification: Verifies the capacity of the registered ScalerNode.
Daemon installation: Installs the necessary Daemon software on the ScalerNode.
Cluster assignment: Assigns the ScalerNode to a specific cluster in the region, based on its hardware class.
Once verified and activated on-chain, the ScalerNode begins earning capacity rewards, and becomes available to Builders for deployment. The deployed Daemon continuously runs challenges against the HyperNode Network (see HyperNode Network section for details) and reports relevant telemetry data. This information is accessible to Hardware Providers via the Console, enabling real-time monitoring and troubleshooting capabilities.