I've never seen them turned on, except at the end of an extending troubleshooting session where a memory limit turned out to be the problem. I can't think of any use cases where it would be beneficial. Why not just assign less memory?
closed as not constructive by RobM, SvW, Magellan, Scott Pack, Zoredache Oct 23 '12 at 16:49
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.If this question can be reworded to fit the rules in the help center, please edit the question.
You really don't want to use them on individual virtual machines, generally, unless you're testing how an application will perform under significant memory pressure to the host. They're much more useful on resource pools and vApps, if you care about ensuring that certain consumers of the system are only getting the total resources they've paid for but don't actually care how many virtual machines they run within those resuorces.
Memory limits on a single virtual machine are an interesting feature, because they serve more of a political role than a technical role, at least according to the instructors of my VCP class a few years back.
Let's say that you work for an IT department that is working on virtualizing all of its older servers. You do an analysis and decide to virtualize some server from your company's Accounting department. Your analysis tells you that even though the server has 32 GB of RAM, it's only ever used 4 GB of RAM at a time, and that was a small load spike. So, you virtualize the system, and to be generous, you give it an 8 GB memory allocation. You don't want to go any higher, because VMware Tools will only balloon out a certain amount of the guest's memory in response to contention and you don't want to go overboard with wasted memory.
That's great and all, but the head of the Accounting department is flippant because his department's budget paid for that server five years ago and damn it, he's gonna get his 32 GB of memory and no, he's not gonna pay for it again because he already paid for it and now you're rambling something about power and cooling and TCO and he doesn't care. Great, chief. Here, check top. See? You've got 32 GB. Great. Leave us alone.
If you're referring to resource reservations, then their value is not when they're assigned per-VM, but rather when they're assigned per resource pool. Overcommitting isn't rare. Using reservations on resource pools helps ensure that tier 1 apps get first dibs at the resources.