Generating a globally unique identifier in Java

Question

Summary: I'm developing a persistent Java web application, and I need to make sure that all resources I persist have globally unique identifiers to prevent duplicates.

The Fine Print:

1. I'm not using an RDBMS, so I don't have any fancy sequence generators (such as the one provided by Oracle)
2. I'd like it to be fast, preferably all in memory - I'd rather not have to open up a file and increment some value
3. It needs to be thread safe (I'm anticipating that only one JVM at a time will need to generate IDs)
4. There needs to be consistency across instantiations of the JVM. If the server shuts down and starts up, the ID generator shouldn't re-generate the same IDs it generated in previous instantiations (or at least the chance has to be really, really slim - I anticipate many millions of presisted resources)
5. I have seen the examples in the EJB unique ID pattern article. They won't work for me (I'd rather not rely solely on System.currentTimeMillis() because we'll be persisting multiple resources per millisecond).
6. I have looked at the answers proposed in this question. My concern about them is, what is the chance that I will get a duplicate ID over time? I'm intrigued by the suggestion to use java.util.UUID for a UUID, but again, the chances of a duplicate need to be infinitesimally small.
7. I'm using JDK6

Answer 1

Pretty sure UUIDs are "good enough". There are 340,282,366,920,938,463,463,374,607,431,770,000,000 UUIDs available.

http://www.wilybeagle.com/guid_store/guid_explain.htm

"To put these numbers into perspective, one's annual risk of being hit by a meteorite is estimated to be one chance in 17 billion, that means the probability is about 0.00000000006 (6 × 10−11), equivalent to the odds of creating a few tens of trillions of UUIDs in a year and having one duplicate. In other words, only after generating 1 billion UUIDs every second for the next 100 years, the probability of creating just one duplicate would be about 50%. The probability of one duplicate would be about 50% if every person on earth owns 600 million UUIDs"

http://en.wikipedia.org/wiki/Universally_Unique_Identifier