Larger customers have unique workflows built into their implementation of Zendesk products. This customization can have drastic impacts on system reliability and performance that isn’t always realized when observed from a typical SLO.
Additionally using aggregate measurements such as percentiles to quantify service performance can have limitations at scale. To deliver a world class customer experience for our customers, we adopted a methodology of measuring request performance using latency bands and individualizing those measurements with customer specific tags.
This presentation shows the technical details of the latency bands that we developed for measuring request performance, and the challenges of implementing those on a per customer basis with metric tags.
Speakers: Jacob D'Onofrio, Fred Moyer
Live Q&A: [ Ссылка ]
