Hacker News Reader

These types of mitigations have the biggest benefit when resources are shared. Do you really think cloud vendors want to lose performance to CPU or other mitigations when they could literally sell those resources to customers instead?

View on HN

They don't lose anything since they sell the same instance which performs less with the mitigations on. Customers are paying because they need more instances.

I imagine they're unable to squeeze as many instances onto their giant computers, though.

There are 3-4 year old servers with slower/fewer cores still operating fine and newer servers operating as well. The generation improvements seem to outweigh a lot of the mitigations in question, not to mention higher levels of parallel work.

Every CPU that isn’t pegged at 100% all the time is leaving money on the table. Some physical CPU capacity is reserved, some virtual CPU capacity is reserved, the rest goes to ultra-high-margin elastic compute that isn’t sold to you as a physical or virtual CPU. They sell it to you as “serverless,” it prints cash, and it absolutely depends on juicing every % of performance out of the chips.

edit: “burstable” CPUs are a fourth category relying on overselling the same virtual CPU while intelligently distributing workloads to keep them at 100%.

Reply on news.ycombinator.com