Amazon EC2 - Do two large instances running in parallel perform (more or less) the same as an extra large instance?

Question

Do two large instances perform the same as an extra large instance? I'm not too familiar with virtualization but just wanted to know from a high level if they are more or less the same.

• Large Instance 7.5 GB of memory, 4 EC2 Compute Units (2 virtual cores with 2 EC2 Compute Units each), 850 GB of local instance storage, 64-bit platform
• Extra Large Instance 15 GB of memory, 8 EC2 Compute Units (4 virtual cores with 2 EC2 Compute Units each), 1690 GB of local instance storage, 64-bit platform

The RAM, # of virtual processors and storage is doubled for an extra large instance. However the price is doubled too. I was under the impression that Amazon would try to incentivize customers to get an extra large instance by making it's price less than two large instances.

Answer 1

Some load is not going to scale out well or only with considerable amount of overhead. So a single, more powerful machine might make more sense than two smaller ones for certain scenarios. In general, some kind of overhead is always there when using more than a single instance:

• since you have more than one operating system instance running, the memory requirements for the kernel and necessary userspace code are multiplied
• the same logic applies for storage - the OS data needs to be stored multiple times
• interrupts (especially periodic ones like timers) need to be processed on all machines independently, using a multitude of CPU cycles compared to a single-instance scenario
• if servers are working on the same data set, cached data is likely to have duplicates among several systems

But then, a scale-out approach is often taken to increase availability. Obviously, all of the above still applies there, but is accepted as the necessary price for independently running system units.