MayaScale on GCP Marketplace

Deploy a 2-node MayaScale active-active storage cluster from the GCP Marketplace. MayaScale always deploys two nodes with server-side replication and automatic failover.

Prerequisites

A GCP project with billing enabled
Compute Engine API enabled
IAM permissions: Compute Admin, Storage Admin, Network Admin, Service Account User

Step 1: Find MayaScale in GCP Marketplace

Go to GCP Marketplace
Search for MayaScale Composable Storage
Click Deploy

Step 2: Configure Deployment

Deployment Configuration

Field	Description	Default
Deployment Name	Unique cluster identifier (lowercase, numbers, hyphens)	—
Performance & Availability Policy	Determines instance type, SSD count, and network config	Zonal Medium

The performance policy is the key decision. It auto-configures everything else:

Zonal Policies (Same-Zone, <1ms Latency)

Policy	Write/Read IOPS	Machine Type	SSDs/Node	Capacity/Node
Zonal Basic	75K / 100K	n2-highcpu-4	1	375 GB
Zonal Standard	130K / 380K	n2-highcpu-8	2	750 GB
Zonal Medium	200K / 700K	n2-highcpu-16	4	1.5 TB
Zonal High	350K / 1.2M	n2-highcpu-32	8	3 TB
Zonal Ultra	800K / 2M	n2-highcpu-64	16	6 TB

Regional Policies (Cross-Zone HA, <2ms Latency)

Policy	Write/Read IOPS	Machine Type	SSDs/Node	Capacity/Node
Regional Basic	63K / 130K	n2-highcpu-4	1	375 GB
Regional Standard	126K / 400K	n2-highcpu-8	2	750 GB
Regional Medium	207K / 780K	n2-highcpu-16	4	1.5 TB
Regional High	333K / 1.2M	n2-highcpu-32	8	3 TB
Regional Ultra	765K / 2M	n2-highcpu-64	16	6 TB

Choose Zonal for maximum performance, Regional for zone-failure resilience.

Compute Resources

Field	Description	Default
Primary Zone	Zone for primary node (secondary auto-selected)	—
Machine Type	Leave empty for auto-selection from policy	(automatic)

Network & Access

Field	Description	Default
Enable Web UI Public Access (Port 2020)	Opens firewall for management UI	false

Step 3: Deploy

Click Deploy. Infrastructure Manager provisions all resources in 3–5 minutes.

Post-Deployment

What Gets Deployed

2 Compute Engine instances with local NVMe SSDs
Dedicated backend network (10.200.0.0/24) with MTU 8896 jumbo frames
Service account with Compute/Storage/Network Admin roles
Firewall rules (SSH, backend replication, optional Web UI)
Placement policy (zonal) or cross-zone distribution (regional)
TIER_1 networking enabled for 30+ vCPU machines

Retrieve Credentials

The password is shown in deployment outputs. From within the instance:

curl -H "Metadata-Flavor: Google" \
  http://metadata.google.internal/computeMetadata/v1/instance/attributes/mayascale-cloud_user_password

Access the Web UI

# Direct (if enabled)
http://<EXTERNAL_IP>:2020

# SSH tunnel (recommended)
ssh -L 2020:localhost:2020 mayascale@<EXTERNAL_IP>
# Then open http://localhost:2020

SSH Access

ssh mayascale@<EXTERNAL_IP>

Connect NVMe-oF Client

From a client VM in the same VPC:

# Install NVMe CLI
sudo apt install nvme-cli   # Debian/Ubuntu
sudo yum install nvme-cli   # RHEL/Rocky

# Discover available targets
nvme discover -t tcp -a <VIP_ADDRESS> -s 4420

# Connect to volume
nvme connect -t tcp \
  -n nqn.2019-05.com.zettalane:mayascale-data-node-1 \
  -a <VIP_ADDRESS> \
  -s 4420

# Verify connection
nvme list
lsblk

The volume appears as /dev/nvmeXn1. Format and mount:

sudo mkfs.ext4 /dev/nvme1n1
sudo mount /dev/nvme1n1 /mnt/data

Multiple Volumes

For policies with 2+ SSDs per node, multiple volumes are available across both VIPs:

# Volume 1 via primary VIP
nvme connect -t tcp -n nqn.2019-05.com.zettalane:mayascale-data-node-1 -a <VIP1> -s 4420

# Volume 2 via secondary VIP
nvme connect -t tcp -n nqn.2019-05.com.zettalane:mayascale-data-node-2 -a <VIP2> -s 4421

Troubleshooting

# Check startup logs (primary node only has startup script)
tail -f /var/log/syslog | grep mayascale

# Check NVMe-oF target status
nvmet status

# Check cluster health
ssh mayascale@<NODE1_IP> "mayacli cluster status"