The SO_REUSEPORT socket option

Posted Mar 28, 2013 6:04 UTC (Thu) by jpb (guest, #87584)
Parent article: The SO_REUSEPORT socket option

Related to "The second of the traditional approaches [...] wake-ups are not fair"

I am just wondering does that really matter? I mean all it matters is that whoever as CPU cycles should pick up the next workload. Now I am not sure why the thread itself picking up the load is actually relevant. The CPU being underutilized seems to solve by itself. I mean if a core is saturated then effectively an extra thread will pick up the next connection and this should work well with what the CPU scheduler is already doing for balancing.

The SO_REUSEPORT socket option

Posted Feb 8, 2014 18:23 UTC (Sat) by batth.maninder@gmail.com (guest, #81449) [Link]

The author seems to be talking in the context of multi-core machines. Imagine a thread pinned to each core. It sounds like author is saying that load is improperly balanced, hence one core would be at 100% and others maybe at 30%, leading to under utilization of cores.