Question 1

How much can we save by migrating from Google Cloud to Hetzner?

Accepted Answer

Savings are often drastic, ranging from 60% to 80%. GCP is known for complex pricing (Sustained Use Discounts) and high costs for managed services (Cloud SQL, GKE management fees). Hetzner offers raw performance at a flat monthly price. For compute-intensive workloads, the price-per-core ratio at Hetzner is unbeatable.

Question 2

How do we handle GCP Egress Fees (Data Transfer costs)?

Accepted Answer

This is often the main driver for migration. GCP charges high rates for data leaving their network. Hetzner includes massive traffic quotas (usually 20TB to unlimited) with every server. For media, SaaS, or data-heavy applications, moving to Hetzner effectively eliminates the "Bandwidth Tax."

Question 3

Does Hetzner offer "Preemptible VMs" or Spot Instances?

Accepted Answer

No, and you don't need them. "Preemptible VMs" are used on GCP to save money at the cost of reliability. Hetzner's standard dedicated servers are often cheaper than GCP's Spot Instances, but they run 24/7 with 100% reliability. You get stable performance without the risk of random termination.

Question 4

Are there hidden costs like "Persistent Disk" IOPS fees?

Accepted Answer

No. On GCP, you pay extra for higher IOPS on Persistent Disks. At Hetzner, you get local Datacenter NVMe SSDs included. You get millions of IOPS and maximum throughput at no extra cost, which drastically speeds up databases and CI/CD pipelines.

Question 5

We rely on GKE (Google Kubernetes Engine). Can Hetzner match this?

Accepted Answer

Yes. GKE is convenient, but it hides the hardware. We deploy Managed Kubernetes (RKE2, K3s, or Vanilla) directly on Proxmox nodes at Hetzner. Benefit: Your worker nodes use high-frequency Ryzen/EPYC CPUs instead of throttled cloud vCPUs. Result: Pods often start faster and run more efficiently.

Question 6

How do we handle Auto-Scaling without GKE?

Accepted Answer

On bare metal, we scale "vertically first". Since hardware is cheap, we provision clusters with enough headroom to handle spikes without needing complex node auto-scalers. For extreme scaling needs, we can script Proxmox to add nodes via the Hetzner API, but for 95% of use cases, static high-performance clusters are more stable and cheaper.

Question 7

Can we migrate Google Cloud Functions (Serverless)?

Accepted Answer

Serverless functions are just code in containers. We migrate your Cloud Functions to Knative running on your new Kubernetes cluster, or to simple containerized services. This removes the vendor lock-in and often reduces "cold start" latency because the underlying hardware is dedicated to you.

Question 8

Do you support Terraform for infrastructure management?

Accepted Answer

Absolutely. We use the Proxmox Terraform Provider. Your DevOps team can keep their "Infrastructure as Code" (IaC) workflows. Instead of defining google_compute_instance, you define proxmox_vm_qemu. The logic remains the same.

Question 9

How do we replace Cloud SQL (Postgres/MySQL)?

Accepted Answer

We build optimized Database Clusters directly on Proxmox VMs using NVMe storage. Without the network-attached storage latency of Cloud SQL, queries often run significantly faster. We handle High Availability (HA) using tools like Patroni or CloudNativePG to ensure production-grade reliability.

Question 10

What is the alternative to BigQuery?

Accepted Answer

BigQuery is a proprietary SaaS product. For many analytics workloads, ClickHouse running on Hetzner bare metal is a superior, ultra-fast, and cost-effective alternative. For general warehousing, large PostgreSQL clusters often suffice. If BigQuery is critical, we recommend a Hybrid setup (App at Hetzner, Analytics on GCP).

Question 11

How do we migrate Google Cloud Storage (GCS) buckets?

Accepted Answer

We deploy MinIO clusters on Proxmox or use Hetzner Storage Boxes. Both offer S3-compatible APIs. We use tools like rclone to sync your GCS buckets to the new storage. Your application simply needs a config update to point to the new endpoint.

Question 12

How do we replace the Global VPC?

Accepted Answer

We use the Hetzner vSwitch (Layer 2) to connect servers within a data center privately. For multi-location setups (e.g., Falkenstein <-> Helsinki), we build transparent WireGuard VPN meshes between the Proxmox nodes. This creates a secure, private network overlay that acts just like a VPC.

Question 13

How secure is Hetzner regarding the US Cloud Act?

Accepted Answer

Google is a US company. Hetzner is a German company. By moving to Hetzner and managing the stack via Proxmox, you shield your data from the US Cloud Act. Your data stays physically and legally in the EU (Germany/Finland), which is a massive compliance advantage for European businesses.

Question 14

What replaces Google Cloud Load Balancing (GCLB)?

Accepted Answer

We deploy dedicated Load Balancers (like HAProxy or Traefik) as entry points to your cluster, or use Hetzner's Managed Load Balancers. For global Anycast needs, we recommend placing Cloudflare in front of your Hetzner infrastructure.

Question 15

How is monitoring handled without Stackdriver/Cloud Operations?

Accepted Answer

We deploy the LGTM Stack (Loki, Grafana, Tempo, Mimir/Prometheus). This open-source stack is the industry standard and offers deeper insights into your hardware and applications than Stackdriver, without the expensive ingestion fees for custom metrics.

Question 16

How long does a GCP-to-Hetzner migration take?

Accepted Answer

Since GCP workloads are often already containerized (Docker/K8s), migration is usually faster than from legacy VMs. Typical projects take 2 to 5 weeks. We sync data in the background and perform a swift cutover.

Question 17

Do I need to manage the Kubernetes cluster myself?

Accepted Answer

Not if you choose Managed Ops. OutaCloud takes the role of your "Platform Engineering Team". We manage Proxmox, the Kubernetes Nodes, upgrades, and security. Your team just deploys kubectl apply.

Question 18

What happens if a Hetzner server fails?

Accepted Answer

We configure Proxmox HA. If a physical node fails, the VMs (and Kubernetes Nodes) automatically restart on the remaining servers. Hetzner support typically replaces failed hardware components within minutes to hours 24/7.

From GCP to Hetzner

Why Hetzner for Your GCP Exit?

Up to 80% Cost Reduction

Full GDPR Compliance

Dedicated Performance

GCP vs. Hetzner + Proxmox

Your GCP to Hetzner Migration

GCP Audit

Architecture Design

Data & Workload Sync

Final Cutover

Choose Your Operating Model

Migration & Handover

Migration & Managed Ops

GCP to Hetzner Migration FAQ

Consider Other Providers

Request Your Individual Exit Plan

Choose Operating Model