Capacity overview

This document provides an overview of how to obtain capacity and quota for each consumption option in Cluster Director.

Cluster Director supports several consumption options to help you get and use compute resources. When you understand the quota, reservation, and billing requirements for each option, you can plan capacity and avoid provisioning delays when you create clusters.

How to obtain capacity

The following table describes how you can obtain capacity for each consumption option:

Consumption option Process for obtaining capacity Pricing implications
Future reservations for capacity blocks

To get future reservation resources, the process is as follows:

  1. To reserve capacity, contact your account team. You can request compute instances and, optionally, a Hyperdisk pool.
  2. Google increases your quota, and then provisions your requested resources at your specified start date and for your specified duration.
  3. When you create a cluster from a template or as a custom cluster, select the Use reservation consumption option and then specify your reservation. Cluster Director creates the cluster from your reserved capacity.
 You pay for the entire reservation period, whether you use the reserved resources or not. For more information, see the pricing for reservations and Hyperdisk pools.
Future reservations in calendar mode

To get future reservation resources, the process is as follows:

  1. Search for available capacity and reserve resources. For instructions, see Create future reservation requests in calendar mode.
  2. No quota is charged and no action is required from you.
  3. When you create a cluster from a template or as a custom cluster, select the Use reservation consumption option and then specify your reservation. Cluster Director creates the cluster from your reserved capacity.
 You pay for the entire reservation period, whether you use the reserved resources or not. For more information, see Dynamic Workload Scheduler pricing.
Future reservations

Future reservations are available only for A3 Mega and N2 machine types. To get future reservation resources, the process is as follows:

  1. Request standard CPU quota or GPU quota for the machine type that you want to use.
  2. Create a specifically targeted future reservation request for A3 Mega or N2 VMs. If you want to submit a future reservation request for A3 Mega VMs, then first contact your account team or the sales team to discuss your request. For instructions, see Create future reservation requests for a single project or Create shared future reservation requests.
  3. After the reservation reaches its start time, create a cluster from a template or as a custom cluster. Select the Use reservation consumption option and specify your auto-created reservation. Cluster Director creates the cluster from your reserved capacity.
 You pay standard rates for the entire reservation period, whether you use the reserved resources or not. You can attach committed use discounts (CUDs) for savings. For more information, see reservation billing.
On-demand reservations

On-demand reservations are available only for A3 Mega and N2 machine types. To get on-demand reservation resources, the process is as follows:

  1. Request standard CPU quota or GPU quota for the machine type that you want to use.
  2. Create a specifically targeted on-demand reservation for A3 Mega or N2 VMs by using the Google Cloud console, gcloud CLI, or Compute Engine API. When you create a single-project reservation that isn't attached to a commitment, you can optionally specify a compact placement policy. For instructions, see Create a reservation for a single project or Create a shared reservation.
  3. When you create a cluster from a template or as a custom cluster, select the Use reservation consumption option and then specify your reservation. Cluster Director creates the cluster from your reserved capacity.
 You pay standard rates for the entire reservation period, whether you use the reserved resources or not. You can attach CUDs for savings. For more information, see reservation billing.
Flex-start

For this consumption option, no reservation is required. To get Flex-start resources, the process is as follows:

  1. Request preemptible GPU quota or TPU quota for the machine type that you want to use.
  2. When you create a cluster from a template or as a custom cluster, select the Use Flex-start consumption option. When your requested capacity becomes available, Cluster Director provisions your cluster. You obtain resources for up to seven days.
 You're charged when the resources are in use. Resources that you provision by using Flex-start receive discounts through Dynamic Workload Scheduler pricing.
Spot

For this consumption option, no reservation is required. To get Spot resources, the process is as follows:

  1. Request preemptible CPU quota, GPU quota, or TPU quota for the machine type that you want to use.
  2. For GPU or CPU instances, view resource availability in the region or zone where you want to create compute instances.
  3. When you create a cluster from a template or as a custom cluster, select the Use Spot consumption option. Cluster Director provisions your cluster, but Compute Engine can preempt the Spot VMs in your cluster at any time.
 You're charged when the resources are in use. Spot VMs receive discounts through Spot VMs pricing.
On-demand

For this consumption option, no reservation is required. To get on-demand resources, the process is as follows:

  1. If you don't have enough CPU quota, GPU quota, or TPU quota for the resources that you plan to provision, then request a quota adjustment.
  2. When you create a cluster from a template or as a custom cluster, select the Use on-demand consumption option. Cluster Director provisions your cluster based on the real-time availability of hardware.
 You're charged when the resources are in use. For more information, see VM instance pricing and Cloud TPU pricing.