$ Why Your Internal Developer Portal
Is a Backstage Graveyard
dispatched: 2026-02-26 · 6 min read · issue-52
Every platform team installs Backstage. Eighteen months later, it has fourteen plugins, zero active contributors, and an on-call rotation to keep it running. This issue dissects what went wrong and what the teams who got it right did differently.
staff engineers · devops leads · sres
## What Ships
Each issue is a deployment log for your brain. Four sections, structured like an RFC. No sponsored content. No vendor press releases.
One real postmortem, anonymized. What failed, what the runbook missed, what got added.
An ADR-style breakdown of a real platform choice. Trade-offs surfaced, not buried.
One tool per issue, benchmarked against alternatives you actually use.
Three links curated from 200+ sources. If it made the cut, it mattered.
52 issues in archive. L0 subscribers access the last 12. L1 gets the full corpus + search.
## Signal-to-Noise Ratio
Your attention is a production resource. We don't waste it.
Scaffold is the only newsletter I forward to my entire platform team. Issue #44 on Karpenter cost modeling saved us ~$140k ARR in cloud spend. The math was right there, reproducible.
We were voluntold into building an IDP after our Heroku migration. Scaffold's golden path series was literally the runbook we didn't have. Three months later we have 40 teams on it.
The ADR breakdowns are the best part. You see the reasoning, not just the outcome. I've referenced the Istio vs Cilium issue in three architecture reviews.
## Stack We Cover
We follow the platform engineering toolchain, not the hype cycle. If it's in your platform-team/ repo, we've written about it.
Coverage does not imply endorsement. We write critically. If a tool has known failure modes at scale, the issue says so by name.
## Community Access Tiers
L0 is the newsletter. L1 is the team behind you when you're the only one in the room who knows what a service mesh is.
hitting a weird latency spike in the sidecar after upgrading istio to 1.20 — p99 goes from 8ms to 340ms under load. anyone seen this?
yes — this is the envoy access log sync issue. disable `--log-format-escaping=json` in your proxy config. we hit this exact pattern in 1.19 too.
holy shit. p99 back to 6ms. that was it. adding this to our upgrade runbook right now.
there's an istio issue thread on this — #47821. not fixed until 1.21. workaround is stable.
Upgrade to L1
- ✓Private Slack — 400+ platform engineers
- ✓Architecture review threads (async)
- ✓Monthly AMA with Staff Eng guests
- ✓Full issue archive + search
- ✓Incident debrief deep-dives