Talk:Memory overcommitment

Generalize

@Ironholds: Memory overcommit is a much more general concept in operating system memory management and isn't necessarily tied to virtualization. -- intgr [talk] 11:38, 20 April 2015 (UTC)

@Intgr: if you can give me some example sources I'm happy to weave those in; I operated from the sources available to me. Ironholds (talk) 14:29, 20 April 2015 (UTC)

@Ironholds: Googling around, it does indeed appear that the virtualization community has largely hijacked this term because it never was a "hot topic" before. But here are some sources I could find quickly:

But the fact that Linux (and other operating systems in some cases) may hand out more memory than they actually have, is a well known fact.

From a cursory reading, there appears to be a significant difference in how the terms are used. In OS terminology, "overcommitting" is when the OS allows applications to allocate more memory than RAM+swap space. But in virtualization, exceeding physical RAM is already considered overcommitting. -- intgr [talk] 15:18, 20 April 2015 (UTC)

Sweet; thanks! I'll take a look at those this evening :). Ironholds (talk) 17:38, 20 April 2015 (UTC)

Windows Overcommit

The article claims that "Windows NT contained overcommitment features" . Is there a source for this claim? I'm very interested in this, since I thought that at least modern Windows don't actually overcommit memory (as in allowing programs to allocate more RAM than the total size of available swap + RAM). Avl (talk) 19:54, 29 March 2019 (UTC)

I believe this might be an issue with the article itself. It may suggest that Windows can "overcommit" in the same way how Linux does (which is not true). The referenced source says "Memory overcommitment is not a feature in itself, but a collection of technologies" which in this case refers to "transparent page sharing" . SleeptightAnsiC (talk) 21:51, 19 March 2025 (UTC)

More about (dis)advantages, criticism and reasoning behind Memory Overcommitment

I've stumbled upon many blog posts criticizing Memory overcommitment and OOM Killer mechanisms in the Linux Kernel, saying that those make System unreliable/unpredictable in (near) OOM situations, and encourage for writing User Space Applications that never expect said condition. Examples: , , , , .

In the other hand, as an advantage, a lot of people claim that it helps a lot for implementing efficient Fork (system call) which the article does not mention anywhere. Examples: ,

Perhaps some of this information can be added to the article? (I do understand those references are not really viable for Wikipedia citation though). SleeptightAnsiC (talk) 21:37, 19 March 2025 (UTC)