What is wrong with Yahoo! JAPAN Tech Blog's "Consideration of Distributed Programming Models and Design Patterns"?

Rather than being wrong, I thought that the overall usage of the term was ambiguous and I didn't understand the meaning well.So it's hard to point out exactly what's wrong, but there's something that clearly says something strange.

Strictly speaking, the servers that make up the cloud environment are MIMD internally for multiprocessors and multicores, and some kind of SIMD is supported on recent expansion boards such as processors and GPUs.Therefore, in strict classification, the current cloud environment is a hybrid architecture of MIMD and SIMD, but from this article, we are focusing on MIMD instead of discussing SIMD trends.

↓ The first half is about the hardware architecture of the servers that make up the cloud.

Strictly speaking, the servers that make up the cloud are MIMD internally for multiprocessors and multicores, and some kind of SIMD is supported on recent expansion boards such as processors and GPUs.

↓ However, for some reason, the second half is about the architecture of the cloud environment.

Therefore, in strict classification, the current cloud environment is a hybrid architecture of MIMD and SIMD, but from this article, we are focusing on MIMD instead of discussing SIMD trends.

The difference between MIMD and SIMD CPU instructions has nothing to do with process-level processing and has no effect on the structure of a cloud environment that uses that process as a unit of processing.

"The ""Flynn's Law"" explained at the beginning is usually literally ""instruction"" and, in other words, ""four-principle operation"" level as shown in the figure."Because distributed systems are by definition MIMD, they are usually used inside "computers" and specifically for "CPU" stories.

On the other hand, in the latter half of the article, "Design Pattern" explains that each "process" has a certain percentage.This will be at the level of "computer" or "process" or "thread."

I don't think each explanation is very wrong, but the article in question should be about the level of the process or computer.I feel uncomfortable because I'm explaining it by jumping from hardware such as MIMD and UMA.

The definition of a design pattern, especially in the design, is defined to be re-used under a name based on past experiences.Therefore, the design pattern itself is easily criticized, and in all cases, silver bullets are not enough.
Now, this article doesn't have a good idea of sharing the issues behind it, and it has this kind of structure because it's hard to understand (like this) and it's hard to find enough examples of distributed programming.

As the questioner said, there are critical articles everywhere, so I hope you will read them carefully before judging them, but I hope you will also consider that writing articles that satisfy all readers is very difficult.