Ameblo, used by Maria Kaze, was renewed from July 1 over 2 days and nights. It was supposedly for performance improvements, but in fact it became far worse, and it can barely be accessed now.
Recently I had just been surprised to see Company S fail by releasing without doing any sizing at all: “There are places like that too!” Ameblo is much the same, making me wonder whether that is simply the state of the world. It is not protected, and I do not think it is that difficult to conduct sizing experiments…
I think the servers are Sun. Perhaps they had the mistaken idea that performance would improve simply by installing a large Sun machine? Of course, I may be wrong…
In any case, this is about sizing experiments. Certainly, if one uses LoadRunner, it may be difficult to reproduce traffic from tens of thousands of people. A LoadRunner license alone can be quite expensive, and preparing machines to generate load may also be difficult. In such cases, apply a separate base load. At my company, we conduct tests using a machine capable of 1 million concurrent sessions and 20,000 new connections per second. Quite a few places use similar machines, but apparently Ameblo did not.
Perhaps I should offer to consult for them, honestly.