HPA Scales to Zero in Kubernetes v1.37: The End of the Always-On Replica (and the One Piece You Still Have to Build)
Kubernetes 1.37 graduates HPA scale-to-zero to beta, so idle tenant services can drop their last replica with no add-on — but HTTP workloads still need an activator. The worked cost math and a git-push PaaS adoption playbook.
Self-Hosted PaaS Showdown 2026: What Coolify, Dokploy, CapRover, and Deploynix Cost When You Need a Second Machine
Adding a second server costs about $12/month on Coolify, Dokploy, or CapRover and $24–$36 on Deploynix Starter — but every panel leaves your database on one disk. A mechanism-by-mechanism walkthrough of each tool's multi-server story and the four problems server two never fixes.
MCP at 97 Million Downloads: the Protocol Won, but Sessions and Sandboxes Are Still Open Problems
MCP hit 97 million monthly downloads and 10,000+ production servers — but stateful sessions still break behind load balancers and every agent framework ships its own sandbox. What the adoption numbers prove, and the two gaps they don't.
CPU Lies: What a Git-Push PaaS Should Autoscale Tenants On Instead
CPU sits at 35% while your service melts — because it was waiting, not working. A concrete pipeline for autoscaling tenants on in-flight requests and queue depth, the adapter-vs-KEDA decision, multi-tenant guardrails, and what Render and Railway get right.
Karpenter Still Doesn't Run on Bare Metal: What vCluster's Auto Nodes Workaround Really Buys You
Karpenter provisions right-sized cloud nodes in under a minute, but it still has no bare-metal provider in 2026. A stage-by-stage latency budget shows what vCluster's Terraform-backed Auto Nodes actually recovers — and why Cluster Autoscaler remains the sane default for a Cluster API fleet on owned hardware.
KEDA's Queue-Depth Autoscaling vs. Render and Railway's Black-Box Scaling
A practical comparison of KEDA's queue-aware Kubernetes autoscaling with Render and Railway's documented CPU, memory, and replica controls, including a worked worker-backlog example.
When Can an OP Stack Batcher Reach $0.001 in L1 DA per Transaction?
A worked model of OP Stack blob batching, the utilization threshold for $0.001 in L1 data-availability cost per transaction, and the limits behind that headline.
K8gb Hits CNCF Incubating: DNS GSLB for Self-Hosted Clusters Without Route 53
k8gb moved to CNCF Incubating on July 18, 2026 — the first vendor-neutral DNS GSLB to clear the bar. One Gslb CR, two Hetzner regions, and 30-second DNS failover with no Route 53 or Cloudflare LB.
Fork in 3ms: What Kedge's Snapshot Tree Means for the Next Agent Sandbox
Kedge forks a hardware-isolated microVM in 3ms from a warm-pool snapshot tree — 10,000x faster than a Cluster API Machine on Hetzner. How the architecture works, what it costs, and what a self-hosted fleet must build to match it.