Why are you surprised? LAMP stack, only at this scale is not a bottleneck. Who knows, maybe they are not even using HHVM, yet. I would be blaming actual code, not the stack for the current trouble.
P.S. LAMP != shitty developers automatically
P.S. LAMP != shitty developers automatically
However Slack seems to be really unstable for being an enterprise product. Communication is a backbone of any business and should therefor be rock-solid. 5000 user maximum? Seriously?