With one hit for a read operation, no matter how many nodes within the cluster (i.e. total volume) and two for a write operation (synchronous write operations) the solutions predictability for the I/O is humongous. The consistency in latency expectations (much lower than any similar product out their) over a huge range of data volume allows for a great development experience.
When one looks at the technology and the algorithmic implementation int he product it looks like a another behemoth of a tech-startup in the making. I think I had read about Larry Page talking about Silicon valley not solving complex technical problems, I guess he missed Aerospike ;-)
On a serious I see this in-memory NoSQL database building up a huge fan base due to the range of application segments that it can cover, adtech, e/m-commerce, fintech, Telco, etc.
Check it out, build on it, if you see something that you don't like or don't understand, hit them with questions. The tech leads are waiting to answer.