This document describes the key features and use cases of Cluster Director,
which lets you deploy, manage, and monitor clusters for artificial intelligence
(AI), machine learning (ML), or high performance computing (HPC) workloads.
Cluster Director is a Google Cloud product that automates the complex
setup and configuration of clusters, helping you configure compute, networking,
and storage resources for your clusters to maximize performance and minimize
downtime.

Cluster Director is designed for IT administrators and AI researchers who
want to avoid the overhead of managing a cluster, and instead focus on running
their workloads.

## Cluster Director key features

Cluster Director integrates multiple Google Cloud services into a
cohesive system that simplifies cluster management and provides the following
features:

- **Automated lifecycle management for clusters**: the Cluster Director
  platform is built around a unified control plane with its own API and UI in
  the Google Cloud console. This feature is a central place to manage the
  lifecycle of your clusters, letting you deploy, manage, and scale clusters
  with all their necessary compute, networking, and storage resources.

- **Flexible cluster orchestration and scaling** : Cluster Director
  supports both a
  [managed Slurm environment](https://docs.cloud.google.com/cluster-director/docs/slurm-orchestration.md.txt) for
  job scheduling and
  [managed compute clusters](https://docs.cloud.google.com/cluster-director/docs/managed-compute-orchestration.md.txt)
  backed by Compute Engine managed instance groups (MIGs). You can use Slurm
  for automated job queuing and dynamic scaling, or use managed compute
  clusters to run workloads on MIGs or integrate your own orchestrator.

- **Resilient, performance-optimized infrastructure**: Cluster Director
  builds clusters on performance-optimized hardware, including CPU-based,
  GPU-accelerated, and TPU-accelerated machines. The system uses
  topology-aware scheduling to colocate compute instances and minimize network
  latency. Cluster Director also provides built-in resiliency, with
  autohealing capabilities that detect and replace unhealthy compute instances
  to help minimize downtimes.

- **Integrated observability**: Cluster Director provides built-in tools
  for monitoring the health, topology, and performance of your clusters
  through a dashboard in the Google Cloud console. These tools let you view
  the health of the components in your cluster.

## Cluster Director use cases

Cluster Director is designed to meet the needs of the following use cases:

| Use case | Example workloads |
|---|---|
| Large-scale AI workloads | - Generative AI and large language model (LLM) training - Fraud detection - Recommendation models |
| HPC workloads | - Complex simulations - Drug discovery, protein folding, and genomic analysis - Risk analysis and quantitative trading |

## What's next

- Learn about the supported machine series for Cluster Director:

  - [CPU-based machine series](https://docs.cloud.google.com/cluster-director/docs/cpu-machines.md.txt)

  - [GPU-accelerated machine series](https://docs.cloud.google.com/cluster-director/docs/gpu-machines.md.txt)

  - [TPU-accelerated machine series](https://docs.cloud.google.com/cluster-director/docs/tpu-machines.md.txt)

- [Choose an orchestrator](https://docs.cloud.google.com/cluster-director/docs/orchestration.md.txt)

- [Try out Cluster Director](https://docs.cloud.google.com/cluster-director/docs/create/quickstart.md.txt)

- [Cluster creation process overview](https://docs.cloud.google.com/cluster-director/docs/create/process-overview.md.txt)