Here you go > Sentry processes 594 trillion events a year across 200,000 organizations.[1] That translates to an average of 18.8M events/sec , peak is likely

Here you go > Sentry processes 594 trillion events a year across 200,000 organizations.[1] That translates to an average of 18.8M events/sec , peak is likely 3-5x of that rather than evenly distributed 365dx24h. Even in 2019/2020 they were doing 500B events a year[2]. There are few single applications anywhere close to that scale, (it is nature of the error monitoring stack), certainly not outside Hyperscalers -Google's Zanzibar [3] comes to mind as definitely larger and equally complicated to be precise as its surface involves security. The parent poster misunderstood my statement, I only said Lago is likely more self-hosted than cloud(like Sentry), not that some single's tenant self-hosted instance is larger than Sentry cloud (!). In terms of reliability they need pretty high nines - nobody would accept their error logging infra to go down, not that different from as payment gateway. While payments is more complex for precision and all the banking integrations you need to manage and reconciliation, error monitoring tooling is orders of magnitude more scale buy its nature. [1] https://blog.sentry.io/self-healing-software https://blog.sentry.io/self-healing-software [2] https://sentry.io/about/press-releases/sentry-application-monitoring-achieves-200-increase-in-enterprise-growth/?utm_source=chatgpt.com https://sentry.io/about/press-releases/sentry-application-mo... [3] https://research.google/pubs/zanzibar-googles-consistent-global-authorization-system/ https://research.google/pubs/zanzibar-googles-consistent-glo...