1732 posts tagged with "Self-Hosting"
Running your own PaaS and infrastructure on machines you own
Cilium Tetragon Hits Production Maturity: What Kernel-Level eBPF Enforcement Adds to a Self-Hosted PaaS's Sandbox
Tetragon 1.4 cleared the rough edges on policy authoring in February 2026. Here's what its in-kernel enforcement actually adds on top of a gVisor/Kata sandbox boundary — with a real TracingPolicy that kills an escape attempt before it lands.
Cloud Native Buildpacks Ship SBOMs by Default: What a Build-Time Bill of Materials Gets You That a Scanner Never Can
Cloud Native Buildpacks generate a CycloneDX/SPDX bill of materials as a build-time side effect, populated from dependencies the buildpack actually resolved and installed. A Dockerfile pipeline's nearest equivalent, Docker BuildKit's SBOM attestation, scans the finished image instead — and misses build-stage-only dependencies right when the EU CRA's September 2026 reporting deadline makes that gap expensive.
Cloudflare's $25 Workers for Platforms Fee vs. Your Own Kubernetes Namespaces: The Real Per-Tenant Math
Cloudflare's Workers for Platforms bills $42.80 a month for 500 metered tenants — your own Kubernetes namespaces on owned Hetzner boxes cost $85. Here's the line-by-line math, and the exact traffic threshold where the answer flips.
Cloudflare Workers Bills $51, Vercel Bills $1,640: The Real Math Behind a 32x Edge-Compute Gap
A worked 100M-request comparison using Cloudflare's and Vercel's own 2026 rate cards: Cloudflare Workers lands at $51.40, Vercel's five metered dimensions land at $1,640.28 for the identical workload — and the gap comes apart into a rate premium, extra metered dimensions, and free-vs-billed bandwidth.
CNCF's Score Spec Wants to Be the Portable Workload Manifest: Should a Git-Push PaaS Target score.yaml Instead of Inventing Its Own Config Format
Score's score.yaml promises one workload manifest that runs anywhere via pluggable provisioners. Here's the real schema, what a 2026 field review says it still can't express, and why a Cluster-API-native PaaS should treat it as an import path rather than its core config format.
Only 41% of AI Developers Call Themselves Cloud Native — Here's the Concrete Chain of Concepts That's Keeping the Other 59% Out
CNCF and SlashData found only 41% of professional AI developers self-identify as cloud native. Here's the actual prerequisite chain — Pods to CRDs to reconciliation loops — standing between an ML engineer and a running service, and where a git-push contract collapses it.
Cozystack 1.4 Ships Fractional GPU Sharing — But the Real News Is a Cluster API Bet Built Differently Than bex's
Cozystack 1.4 shipped HAMi-powered fractional GPU sharing and a rewritten dashboard, but the real story is architectural: Kamaji-hosted control planes and KubeVirt VMs give every tenant a nested Kubernetes cluster — a fundamentally heavier Cluster API bet than a flat, container-only design like bex's. Here's what each buys and what it costs.
Depot's Colocated BuildKit Cache: What 2x-55x Faster Builds Actually Require
Depot's 2x-55x faster builds come from one architectural trick — keeping the Docker layer cache permanently colocated with the builder instead of fetching it over the network. Here's how it works, and how to build the same thing on your own Kubernetes fleet.
DigitalOcean Is Deprecating Legacy App Platform Plans With No Fixed Date: The Exact Bill Difference, Tier by Tier
DigitalOcean's App Platform legacy plans — any app created before May 7, 2024 — are being deprecated with no fixed date attached. Here's the exact price and egress math across every tier, and what an undated migration means if you're not watching for it.