It is hard to recommend Google Cloud

Original Article ↗ Hacker News Discussion ↗

Engineers and operators describe Google Cloud as technically strong but strategically unreliable, citing frequent product deprecations, surprise shutdowns (like Google Domains and IoT), and weak or bot-driven support as major trust breakers. Many compare it unfavorably with AWS and, to a lesser extent, Azure, which they see as more stable for long-term services despite worse UX or higher costs. A recurring theme is that Google’s short-term decisions and lack of “enterprise DNA” make it hard to recommend GCP for critical infrastructure, even while some core offerings like Cloud Run, GKE and the CLI are praised.

Customer support and responsiveness

  • Many describe GCP support as slow, indifferent, and disempowered, even for large, high-paying customers.
  • Sales and “contact us” flows are often bot-gated, with difficulty reaching a human.
  • In contrast, AWS is repeatedly cited for responsive support (even for small accounts) and willingness to get engineers on calls; Cloudflare is also praised for quick public responses.

Product churn, deprecations, and migrations

  • Frequent API and product deprecations (Cloud IoT, Container Registry, Photos API changes, Firebase/Android APIs, etc.) are a major complaint.
  • Migrations are seen as poorly documented, labor-intensive, and often unnecessary, draining developer time.
  • Some note that AWS also deprecates services, but usually more cautiously, with longer grace periods and life-support rather than hard shutdowns.

Google Domains, DNS, and optics

  • Shutting down Google Domains is viewed as a self-inflicted trust disaster, especially for such foundational infrastructure.
  • Many moved domains to Cloudflare or Route 53 and now avoid GCP for anything critical.
  • Cloud Domains’ unclear status (limited availability, “deprecated” transfers) deepens concerns.

Trust, bans, and catastrophic failures

  • Several reference Google’s reputation for sudden account bans and opaque, unfixable enforcement.
  • The UniSuper incident (accidental deletion of a private cloud environment) and a Mozilla outage tied to a GCP change are cited as red flags, though details and blame are debated.
  • This leads some to consider GCP a “hard no” for serious business use.

Comparisons with AWS and Azure

  • AWS: seen as the most reliable and “customer-obsessed,” but expensive, bloated, and with complex IAM/UX.
  • Azure: often described as unreliable and confusing, yet strong in enterprise integration and long-term support; many big orgs prefer Microsoft 365/Azure stack.

GCP strengths and weaknesses

  • Many engineers praise GCP’s technical design, UI, CLI, IAM model, logging, Terraform snippets, and products like Cloud Run and GKE.
  • Core services are reported as stable by long-time users.
  • Weaknesses: sluggish console, convoluted permissions, lack of spending caps, surprising pricing (e.g., load balancers, egress), and products that feel under-resourced or abandoned.

Alternatives and cloud skepticism

  • Some advocate self-hosting or simpler VPS/second-tier providers (Hetzner, OVH, IONOS, etc.) for cost, control, and stability.
  • There is broad cynicism toward all big clouds: AWS = cost extraction, Azure = reliability issues, GCP = churn and trust problems.