Dora Noda
Software Engineer
Dora Noda is a Software Engineer writing about Web3 infrastructure, blockchain networks, and self-hosting.
3404 posts · View all authors
Dear Friend, You Have Built a Kubernetes: Why HN Keeps Resurfacing This Essay
A 2024 satirical essay about hand-rolling a worse Kubernetes hit Hacker News again in 2026. Lined up against a real 2025 postmortem, the argument turns out to be a sharper pitch for owning Cluster API than any feature list — with the honest caveat about when it doesn't apply.
Deno Deploy Classic Shuts Down July 20, 2026 — and Takes Subhosting v1 With It
Deploy Classic and Subhosting's v1 API both die July 20, 2026 — the exact migration checklist, what Subhosting's isolation model actually took to build, and why letting a tenant resell tenants is a scope-creep trap for most git-push PaaS platforms.
Devin Retired Its Pricing Unit for Everyone Except the Customers Who Actually Negotiate
Cognition quietly killed Devin's ACU billing for every self-serve customer in April 2026, keeping the unit only for Enterprise contracts — a concrete lesson for how a platform should meter AI-agent-driven work instead of inventing a new billing currency.
Docker Hub's Pro Tier Jumped 80%, Team 67%: The Break-Even Math for Running Your Own Registry
Docker Hub's Pro and Team subscriptions jumped 80% and 67% in a repricing still in effect today. A line-by-line recompute of what that costs a 5- and 20-person team against running Harbor on a single owned Hetzner box, including the operational costs a "just self-host it" pitch usually skips.
Doppler Finally Ships On-Prem — It's Still Not Infisical: What Self-Hosting Secrets Actually Costs
Doppler shipped an on-prem deployment option in June 2026, but it's Enterprise-only and closed-source. Infisical's MIT core stays free to self-host — until you price out RBAC, SSO, and per-identity billing against Doppler's per-seat model.
Dremio's MCP Server Doesn't Give AI Agents a Separate Policy — Here's the Kubernetes Version for a Self-Hosted PaaS's Deploy Tools
Dremio's lakehouse MCP server enforces the same row/column policies for an AI agent as for the human who's logged in — no separate agent policy to write or drift out of sync. Here's the exact Kubernetes primitives (TokenRequest, RBAC impersonation, audit logging) that build the same guarantee into a self-hosted PaaS's deploy and rollback tools.
The 11,000-Hour Myth: E2B's Real Self-Hosted Break-Even on a Hetzner Cluster-API Fleet
The widely-cited 11,000-sandbox-hour break-even for self-hosting E2B-style sandboxes turns out to compare an H100 GPU bill against a CPU-only rate. Here's the real number, recomputed against Hetzner Cluster-API pricing.
EKS 1.33 Loses Standard Support July 29, 2026: What the 6x Extended-Support Markup Really Costs
Amazon EKS quietly enrolls aging Kubernetes 1.33 clusters into extended support at 6x the hourly rate on July 29, 2026 — here's what that markup costs in dollars, and what it actually costs to own the upgrade cadence with Cluster API instead.
Fly.io Kills GPU Hosting on August 1, 2026: Where the Workloads Go, and What It Actually Costs
Fly.io's GPU Machines go dark for good on August 1, 2026. Here's the real migration math — rental marketplaces vs owned Hetzner hardware — and why the shutdown is really a lesson about whose roadmap your infrastructure lives on.