Optimizing the Number of Cores per Socket for VMs

[RenxiaoChen]

Figure ONE

Figure TWO

An article says that  Figure ONE's architecture is more optimized than Figure TOW's,

WHY  ? PLEASE HELP ME . TKS

[daphnissov]

An article says that

In the future when you quote an article and copy graphics from it, please give us the source link for validation.

As a general statement, figure one is *usually* optimal because by configuring a "wide" VM (one in which sockets are used over cores) it allows ESXi to place vCPUs into the optimal NUMA node for scheduling purposes. Things have changed a bit as of 6.5 and 6.7 so, again, this is just a general statement.

[RenxiaoChen]

Sorry, the article is Chinese, so I did not give the source links, and the article is only said that Figure TWO( not ONE . The clerical error above， SORRY!) is  better

So my real question is  :     WHY      Figure TWO's architecture is more optimized than Figure one's,

首先谢谢您的回答。很抱歉，文章是中文的，所以我没有给出URL 链接。而且文章也仅仅是说图二的架构更合理。我就想知道，为什么图二的架构更加合理。

[daphnissov]

Sorry, the article is Chinese, so I did not give the source links

Regardless of the language, please provide source links next time.

the article is only said that Figure TWO( not ONE . The clerical error above， SORRY!) is  better

Without context (which the source would provide), option two is generally not "better" in a generic sense because it makes some assumptions which may not be true. But taking two images potentially out of context from the original article is difficult to make commentary on.