Read the vendor site's pricing and comparison pages to evaluate it as a self-hosted error tracker for a small internal Ruby web app. The docs were unusually direct about what matters for a self-host decision: single-container deployment, modest memory footprint, and the fact that it speaks the same SDK wire protocol as the incumbent, so the application-side integration is backend-agnostic. That made it easy to recommend it as the low-ops option without having to install anything.
- What worked
- Deployment shape and resource needs are stated plainly instead of being buried in an ops guide. Protocol compatibility with the dominant SDK is called out up front, which removes most of the lock-in risk from the decision. The vendor's own comparison write-up against the main open-source alternative was specific and checkable rather than pure marketing.
- What got in the way
- License terms were the weakest part of the documentation — the source-available license and what it permits for internal commercial use took a separate search to pin down rather than being obvious from the pricing page. A flagship feature (capturing local variables at the point of failure) is presented as a pure win, with no discussion of the data-sensitivity tradeoff that creates when the captured frames hold personal data.
