MayaScale Terraform

Terraform modules for deploying MayaScale across GCP, AWS, and Azure. Use these for automated, repeatable deployments with CI/CD integration.

Module Locations

Cloud	Path	Description
GCP	gcp/mayascale/	Direct Terraform module
GCP Marketplace	gcp/mayascale/marketplace-package/	Marketplace-wrapped module
AWS	aws/mayascale/	EC2 with NVMe instance storage
Azure	azure/mayascale/	VMs with local NVMe

GCP

Quick Start

module "mayascale" {
  source = "./gcp/mayascale/marketplace-package"

  project_id              = "my-project"
  goog_cm_deployment_name = "prod-storage"
  zone                    = "us-central1-a"
  performance_policy      = "zonal-medium-performance"
}

Key Variables

Variable	Type	Default	Description
project_id	string	—	GCP project ID
goog_cm_deployment_name	string	—	Cluster name
zone	string	us-central1-a	Primary zone
performance_policy	string	zonal-medium-performance	Performance tier
machine_type	string	""	Override (auto-selected from policy)
enable_web_ui_access	bool	false	Open port 2020
use_spot_vms	bool	false	Spot VMs for testing
client_nvme_port	number	4420	Base NVMe-oF port
client_protocol	string	nvme	nvme, iscsi, or both
client_exports_enabled	bool	true	Auto-create client volumes

Performance Policies

# Zonal (same-zone, <1ms latency)
performance_policy = "zonal-basic-performance"      # 75K/100K IOPS, n2-highcpu-4, 1 SSD
performance_policy = "zonal-standard-performance"   # 130K/380K IOPS, n2-highcpu-8, 2 SSDs
performance_policy = "zonal-medium-performance"     # 200K/700K IOPS, n2-highcpu-16, 4 SSDs
performance_policy = "zonal-high-performance"       # 350K/900K IOPS, n2-highcpu-32, 8 SSDs
performance_policy = "zonal-ultra-performance"      # 800K/2M IOPS, n2-highcpu-64, 16 SSDs

# Regional (cross-zone HA, <2ms latency)
performance_policy = "regional-basic-performance"    # 60K/100K IOPS
performance_policy = "regional-standard-performance" # 120K/380K IOPS
performance_policy = "regional-medium-performance"   # 180K/700K IOPS
performance_policy = "regional-high-performance"     # 315K/900K IOPS
performance_policy = "regional-ultra-performance"    # 720K/1.8M IOPS

Key Outputs

Output	Description
cluster_name	Deployed cluster name
vip_address	Primary VIP for client connections
storage_vip_primary	Primary storage VIP
storage_vip_secondary	Secondary storage VIP
mayascale_password	Web UI password
web_ui_url	Web UI URL
ssh_primary	SSH command for primary node
client_volumes	NVMe-oF volume endpoints (NQN, port, VIP)
nvmeof_port	Base NVMe-oF port
total_storage_gb	Total NVMe capacity
performance_characteristics	IOPS targets, capacity, instance details
network_architecture	Frontend/backend network IPs

What Gets Deployed

• 2 Compute Engine instances with local NVMe SSDs
• Dedicated backend network (10.200.0.0/24, MTU 8896)
• Placement policy (zonal) for collocated VMs
• TIER_1 networking for 30+ vCPU machines
• Service account with required IAM roles
• Client volume NVMe-oF exports (if enabled)

AWS

Quick Start

module "mayascale" {
  source = "./aws/mayascale"

  cluster_name       = "prod-storage"
  performance_policy = "zonal-medium"
  key_pair_name      = "my-key"
  region             = "us-east-1"
}

Key Variables

Variable	Type	Default	Description
cluster_name	string	—	Cluster name
performance_policy	string	zonal-medium	Performance tier
key_pair_name	string	—	SSH key pair
region	string	—	AWS region
use_spot_instances	bool	false	Spot instances
client_nvme_port	number	4420	Base NVMe-oF port
client_exports_enabled	bool	true	Auto-create volumes

Instance Selection

AWS uses NVMe instance storage families:

Tier	Instance Type	NVMe Devices	Capacity
Basic	i4i.xlarge	1	937 GB
Standard	i3en.2xlarge	2	5 TB
High	i3en.6xlarge	4	15 TB
Ultra	i3en.12xlarge	8	30 TB

Key Outputs

Output	Description
vip_address	Primary VIP
client_volumes	NVMe-oF endpoints
ssh_commands	SSH commands for both nodes
deployment_summary	Full deployment details
node1_external_ip / node2_external_ip	Node public IPs

Azure

Quick Start

module "mayascale" {
  source = "./azure/mayascale"

  cluster_name = "prod-storage"
  location     = "eastus"
  vm_size      = "Standard_L8s_v3"
}

Common Workflow

# Initialize
terraform init

# Plan
terraform plan -out=tfplan

# Apply
terraform apply tfplan

# Get client connection info
terraform output client_volumes
terraform output vip_address

# Connect NVMe-oF client
nvme connect -t tcp \
  -n nqn.2019-05.com.zettalane:mayascale-data-node-1 \
  -a $(terraform output -raw vip_address) \
  -s 4420

# Destroy
terraform destroy

Validation

After deployment, use the validation scripts:

# MayaScale validation (includes FIO performance tests)
./common/validate-mayascale.sh --cloud gcp   # or aws, azure

# Check deployment outputs
terraform output deployment_summary