In response to Steeg Martin Dr. <[email protected]>: > Hi there, > I'm already applying Slony-I in a small single-master / single-slave > application test environment (non-commercial) and I have some question > considering the applicability of Slony-I / Slony-II in some more general > situations. > > These questions are: > > Ability to provide WAN replication.
Much of this is covered in the docs: http://slony.info/documentation/slonyintro.html#INTRODUCTION > - Is it possible to use Slony-I in wide-are network (WAN) enviroment ? > The bandwidth on the master-master link is 34 Mbit/sec, on the > master-slave link 10 MBit/sec. Both sides will have a latency of 30msec. Yes. We use it extensively in a WAN environment that frequently has worse latency/bandwidth than that. > Robustness concerning network drop out. > - Can (slave) nodes which are not continuosly connected to network be > used ? Yes, but with caveats. If the network goes down for a short while once a month or so, you'll be fine. If the network is constantly dropping out (like for hours at a time on a daily basis) Slony probably won't work for you. > Satellite Connection. This scenario considers the replication between > master and slave nodes communicating over satellite. > - Is Slony (-I, maybe the sucessor -II) capable for Replication over > satellite network (latency 850msec, bandwith 2 MBit/sec) ? Where are you seeing anything about Slony-II? AFAIK, that project has been dead for years. Whether or not Slony can handle your satellite network is a factor of whether the frequency of your data change exceeds the bandwidth of your connection. 2mb/sec is pretty tight, so I would say "no" if you've got a DB that's being constantly updated, but "yes" if it's largely a read only database that sees only occasional updates. Of course, that's a sliding scale. You'll have to test your particular case to see if it will work under those constraints. > Data consistency. In this scenario, the two master nodes are > unconnected, changes are made and new primary-key values are generated > at both sites. > - How will the upcoming Slony-II act by time the master nodes are > reconnected ? How about conflict resolution in the case of newly > introduced, equal PK-values at each (unconneted) site ? Is there any > design proposal ? Again, I don't think the Slony-II project is viable any more. A google search doesn't turn up anything about it any more. Slony-I is not capable of multi-master replication. -- Bill Moran http://www.potentialtech.com http://people.collaborativefusion.com/~wmoran/ _______________________________________________ Slony1-general mailing list [email protected] http://lists.slony.info/mailman/listinfo/slony1-general
