Distributed methods are difficult. Add networking to the combination, and also you get traumatic challenges just like the CAP theorem and Byzantine fault tolerance. Most of these challenges are unknown to engineers who must undergo via the seller advertising and marketing shows, making it exhausting to find out whether or not the most recent shiny gizmo works exterior of PowerPoint.
I began gathering articles describing distributed-system gotchas years in the past, wrote quite a few weblog posts on the subject within the heydays of the SDN Will Save the World lemming run, and arranged them into the Distributed Methods Sources web page.