From: James F. <jam...@gm...> - 2009-07-07 20:19:43
|
>> I also see an option 3): Implement our own replication mechanism which is >> tailored to eXist. This could support features beyond simple replication, >> e.g. queries against distributed collections (we can still think about >> replacing our btrees). I think option 3 is something to consider as well. jgroup was the result of trying to bolt on clustering ... with limited success. I think that what we are actually talking about is something larger then just replication ... peer to peer may offer some attractive advantages and we could implement our own mapreduce approach ... J |