This may sound like yet another, "what is a vcpu" or "should I use elasticache, remember its single threaded!!" question --- but I think this asks a question unanswered.
redis is single threaded, means only one core will ever be used.
elasticache is managed service for redis. AWS offers multiple instance types for elasticache with varying memory sizes and amounts of vcpus.
AWS is aware that even though the instance has more than one vcpu, performance is capped at only one of those vcpus.
So here's my question: Are all instance types suffering the same performance loss? Naturally, if a computer has a 2 GHz processor, and another computer has a latest gen 3 GHz processor -- the latter would perform better.
However, AWS has its mystery vcpu that doesn't expose its processor model or clock speed. Naturally, the 0.729/hour cache.r3.8xlarge should perform better than the 0.046/hour cache.r4.large. But, despite one having 2 vcpus, and the other having many many more (site doesnt list the number) .. they would both only use 1 vcpu. Which, logically, I would think would be the same performance.
I am unable to believe that a processor supporting 32+ GB memory would perform the same way a processor supporting 2 GB would. At the higher price ranges, I have to believe the processors are much higher grade than the entry level price ranges.
If trying to convert a present instance to elasticache, it would seem that one could match an underpowered inappropriate elasticache. Even though memory was exceeding their requirements. Or the contrary, vastly overpay for an unnecessary cache that again still meets memory requirements.
Is the CPU performance truly the same (somehow) ? If it is, then the super pricey cost of a cache applies only to memory and network; ignoring CPU entirely..
