Re: [infinispan-dev] KeyAffinityService
Sent from my iPhone
On 14 Jun 2013, at 20:02, cotton-ben <ben.cotton(a)ALUMNI.RUTGERS.EDU> wrote:
/
> Just to reiterate, your *never* expects a node to crash? /
Easy now, Mircea!
All of us in the distributed computing community know
I'm just trying to help
Ben.
that we live in a
world of necessary compromises (i.e. Brewer's CAP theorem, etc.) re: limits
of service guarantees and capabilities.
Now, of course, I do expect a node to crash. And I do know that such an
event will have consequences for my beloved pinnedKey set.
But, just as we make compromises because we *definitely* cannot
simultaneously provide CAP (BTW, Prof. Lynch in fact *proved* Brewer's CAP
conjecture) ... it does not mean that we should not 'go for it'.
The 'P' in CAP is for 'Partition healing' capability of a distributed
system. If my node partition crashes, will just have to evolve a 'healing'
strategy. I will be patient here. I have to be.