About Compute Nodes

A compute node provides the ephemeral storage, networking, memory, and processing resources that can be consumed by virtual machine instances.

CloudSystem supports two types of compute nodes:

  • ESX clusters: Clusters are created in VMware vCenter Server. When the vCenter Server is registered and configured in the CloudSystem Console, clusters can be imported from vCenter Server and activated in the cloud.

  • KVM compute nodes: KVM compute nodes are created manually. Once created, the Cloud Management Network enables the CloudSystem Console to see the KVM compute nodes. When the CloudSystem Console detects a compute node, it is automatically added to the Compute Nodes screen in an Unknown status.

Compute nodes in the cloud

Within a cloud environment, compute nodes form a core of resources. They supply the processing, memory, network, and storage that virtual machine instances need. When an instance is created, it is matched to a compute node with the available resources. A compute node can host multiple instances until all of its resources are consumed.

Managing compute nodes

Activating an ESX cluster or a KVM compute node performs the configuration required to bring the system into the cloud. This includes the following actions:

  • Activating an ESX cluster: Verifies virtual switch settings and notifies the Proxy appliance to start managing instances deployed on the cluster.


    [IMPORTANT: ]

    IMPORTANT: Do not activate an ESX cluster in more than one cloud.


  • Activating a KVM compute node: Installs agents, identifies network interfaces, and verifies the operating system and the compute node status.

The Compute Nodes screen in the CloudSystem Console displays all available ESX clusters and KVM compute nodes, along with their resources. You can perform the following tasks from this screen:

  • Import an ESX cluster from vCenter Server.

  • Activate an ESX cluster or a KVM compute node.

  • Deactivate an ESX cluster or a KVM compute node.

  • Delete an inactive, previously imported ESX cluster or expire the DHCP lease on a KVM compute node.

The CloudSystem Portal contains a Hypervisor tab, which allows administrators to see all activated ESX clusters and KVM compute nodes, along with the number of instances attached to each.

Can I delete compute nodes from the cloud?

You can use the delete action to remove an imported ESX cluster, if it has been deactivated. Deleting an inactive ESX cluster removes it from the Compute Nodes overview screen, but does not affect the ESX cluster itself, since it is managed in vCenter Server. See Deactivate a compute node, and Delete a compute node.

When the delete action is used on KVM compute nodes, the DHCP lease is expired and the compute node is removed from the screen.

Understanding compute node data

You can click the icon next to an ESX cluster or KVM compute node on the overview screen to show all available data. See Overview screen details for an explanation of the compute node data.

Two groups of horizontal graphs display when the compute data is expanded. The top set of graphs show the allocation usage of the instances. See Calculating the number of instances that can be provisioned to a compute node. The bottom set of graphs show the physical usage of the instances.

Allocation graphs

These graphs show the virtual size, which is the physical size multiplied by the oversubscription rate. For clusters, this is the aggregate value of all hypervisors in the cluster.

  • CPU allocation: Amount of CPU cores designated for instances

  • Memory allocation: Amount of memory designated for instances

  • Storage allocation: Amount of compute storage designated for instances

Physical usage graphs

These graphs show the physical size information. For clusters, this is the aggregate value of all hypervisors in the cluster.

  • CPU usage: Number of actual CPU cores consumed by instances within the last five minutes

  • Memory usage: Amount of actual memory consumed by instances within the last five minutes

  • Storage usage: Amount of actual compute storage consumed by instances within the last five minutes

You can click the icon to collapse the detailed view.

When one or more virtual machine instances are attached to an ESX cluster or KVM compute node, you can click the instance link next to Contains. The link takes you to the Instances screen, where you can view the Instance details. See Overview screen details for an explanation of instance details.

The Dashboard provides another view of compute node resources. See Interpreting the Dashboard data.

Understanding the data filter

By default, information is listed on the Compute Nodes overview screen in ascending alphabetical order, according to the compute node names. You can use the Sort by options to sort the listed compute nodes by any of the available choices in the drop-down list. For option descriptions, see Overview screen details.

You can click the icon on the overview screen to expand (or collapse) the Filters pane on the left side of the screen.

Use the filter selectors to refine the display of information. 

  • Select a status level to display only compute nodes with a specific status, or leave the default filter setting to display compute nodes in All statuses. See Status and severity icons.

  • Click Reset to restore the filters to the default setting.

Adding compute nodes to the cloud

Compute nodes are added to the cloud in different ways, depending on the type of compute node you want to add.

ESX clusters appear on the Compute Nodes screen after you complete the following two steps:

  • Register the managing VMware vCenter Server. This action provides CloudSystem Foundation with the location and credentials of the vCenter Server managing the ESX cluster.

  • Import a cluster. This action calls vCenter Server and retrieves information about an ESX cluster. The cluster is added to the Compute Nodes overview screen in an Unknown state. The first time a cluster is imported, the proxy appliance establishes communication between vCenter Server and the CloudSystem Console. The proxy appliance can support up to 12 ESX clusters. When more than 12 ESX clusters are imported, a new proxy appliance is spawned.

KVM compute nodes are issued a DHCP lease from the Cloud Management Network, and then they appear on the Compute Nodes screens in an Unknown state.

After an Activate action is performed, ESX clusters and KVM compute nodes are ready to host instances.

Calculating the number of instances that can be provisioned to a compute node

The maximum number of virtual machines that can be provisioned to a compute resource is based on the following:

  • Amount of installed memory, available disk capacity, and number of CPU cores on the compute resource

  • Flavor settings of the virtual machines to be provisioned

  • Resource oversubscription, which is individually applied to the memory, disk, and CPU calculation

Resource oversubscription rates for KVM and ESX shows resource oversubscription rates so you can properly dimension the capacity of your compute resources based on virtual machine size requirements.


[NOTE: ]

NOTE: These rates are hard coded in CloudSystem Foundation and cannot be changed.


Resource oversubscription rates for KVM and ESX

Physical resource Virtual resource Physical to virtual oversubscription rate
1 CPU core 8 CPU cores 1:8
1 GB RAM 1.5 GB RAM 1:1.5
1 TB disk 1 TB disk 1:1

See also