Turns out the YABS is just hard as fuck on processors, which was reflecting running it on another provider in a VM and seeing the node spike even farther than mine. I guess that a spike is not so bad, but I can see how someone who took just the load-y part of YABS and looped it could cause problems for a provider.