Host factory overview

Host factory provisions compute hosts from cloud providers to meet resource demand in your cluster.

When IBM® Spectrum Symphony is set up for cloud bursting, the HostFactory service runs on a management (primary) host in the cluster. This service handles all resource requests to the cloud providers. When resource demand peaks in your cluster, the cluster dynamically extends to include compute hosts from the cloud to satisfy workload demand. You can control bursting for your cluster through policy configurations, which define when and how resource scale-out and scale-in requests are triggered. Host factory can be enabled in an on-premise cluster or in a cluster in the cloud to support various scenarios.

The following image illustrates a few cloud bursting scenarios when host factory is enabled:

Cloud-bursting scenario, showing communication from the primary host in your on-premises cluster to the cloud through a VPN tunnel.
The main components of the host factory framework are the HostFactory service, your cluster's requestors, and cloud providers.
High-level architecture Key components
Architecture of host factory that shows the main components
 HostFactory service: An EGO service that runs on one of the management hosts in the cluster and handles all resource scale-out and scale-in requests to cloud service providers. Only one instance of the HostFactory service exists per cluster.
Requestor instance: An entity that can request cloud resources through host factory. A requestor instance can initiate cloud provisioning manually or dynamically (by using a requestor plug-in). IBM Spectrum Symphony provides the following built-in requestor instances for use within your cluster:
  • The admin requestor instance, which is enabled by default in REST mode and used for manual cloud provisioning from the cluster management console.
  • The symAinst requestor instance, which when enabled in POLL mode triggers cloud bursting dynamically based on application resource demand.
Requestor plug-in: A program that checks cluster workload and triggers scale-out requests for cloud compute hosts or scale-in requests to release cloud hosts. The requestor plug-in is used when a requestor instances is enabled for dynamic provisioning (POLL mode).
Provider instance: An entity that provides hosts to host factory. A provider instance encapsulates cloud account and configuration to be used for provisioning. The provider instance uses a provider plug-in to connect to the cloud provider.
Provider plug-in: A program that uses the cloud provider's APIs to communicate with the cloud provider for provisioning, monitoring, and releasing of cloud hosts. The HostFactory service communicates with the cloud provider through the corresponding provider plug-in provision resources for your cluster.