[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: Multimaster further work



On Wed, 23 Oct 2002, Derek Simkowiak wrote:

[ Interesting proposal ]

> 	There is a chance that the MD5's will be identical for two
> different sets of DN data, but I have read before that the Universe is
> likely to end before that actually happens.  I'm not a hashing expert, so
> I'm not sure.  But we could easily come up with some kind of
> conflict-resolution scheme if that happens, such as adding the originating
> master's I.P. address to the data before hashing.

This will make no difference; you still have an infinitesimal chance of
a collision.  The data is unpredictable to start with; adding something
to it does not make it any more or less predictable.

Personally I think all your servers are more likely to fall over at the
same time before this would happen.

It's a bit like being worried about a shark attack; unless you make a
habit of swimming in shark-infested waters, you're more likely to be
killed by a falling coconut.

-- 
Dave Horsfall  DTM  VK2KFU  daveh@ci.com.au  Ph: +61 2 9906-4333