This is the roadmap for the Speedshop Ruby on Rails performance retainer service. It was last updated March 28th, 2025. Items are in rough priority order.
In-Progress:
- Alerting and dashboard scorecards: We want everyone to be on the same footing with SLOs for queues and tracking our core set of 20 metrics, plus your app's specific key metrics (see next bullet). The output will be a scorecard page that's pass/fail on each point.
- Formalizing key metrics with everyone: Since we started the retainer, we learned it's helpful to have a set of 3 or fewer key metrics to monitor and improve upon over the course of the retainer. If we haven't already formalized these with you, we'll do so.
- Automated checks: jemalloc, YJIT, RMP, SQL comments, bcrypt version. We have a few automated checks that can be implemented just by reading your source code and doing a couple of static checks. We'll also ship some detailed guides for "why" and "how" to implement each.
- Sidekiq/Puma auto-tuner: Yuki continues to work away at adaptive concurrency for Puma and Sidekiq. No more setting thread counts, and automatically adjusts to changing workloads.
Backlog
- A "performance score". This score, similar to Lighthouse, basically audits your team and app on a number of set "checks" (there will probably be 50 in total). Some of this is “are you running with jemalloc”, others might be “response time below subjective Nate thresholds or meeting best practices”.
- Client portal. Uses magic link sign in based on email domain.
- Core Web Vitals monitoring
- Ruby Platform-Kai, a semi-public meetup for Ruby platform teams.
- Quarterly executive briefings on long-term trends, risks, growth opportunities in platform we’re seeing, like a mini-conference.
- Ultra Tier. Aimed specifically at companies of >100 Ruby engineers. Includes additional training, company laptop usage, more.
- Mini Tier. Aimed at companies of 10 devs or less. Limited communication, no sync calls.