Author here! We're pretty keen on fast CI times, from the developer happiness angle to agility when you want to roll-forward a bad deploy. Our ideal is less t

Author here! We're pretty keen on fast CI times, from the developer happiness angle to agility when you want to roll-forward a bad deploy. Our ideal is less than 5m from merge to production, and were unhappy when our pipeline became >10m. It turned out that CircleCI's docker layer caching feature had interacted badly with us hiring more devs (and consequently, shipping more frequently!) so we hosted our own single purpose container builder to get our builds back on track. The article goes into a bit of detail around Docker layer caching and how this works in Circle. Hope you find it interesting!