Re: database distribution/replication

"Jens" <axvkkkclbcqg_at_spammotel.com> wrote in message news:vImTVl25inkn-pn2-ZNdWUY5oEGdd_at_jms.xww.de...
> In several countries we have unix data bases in our own format (thus
> no db/2, oracle or the like), in each case approx. 50'000 of
> sentences.
> These data bases are to be synchronized among each other or to one
> master.
>
> Changes in one country should be done in other countries also, quasi
> live (thus not by night batches). We would like to develop the
> programs for that.

Are they all permanently connected to the network? How many users are there? Do any of the sites 'own' thier own data? Why do offices want to view other offices data, just to read it? Or to write also?

I'm no expert, but there are many different topologies for replication, each being more suitable to different situations. It sounds like your replication might be of a complex nature, so I'd be reluctant to write my own engine when there are several RDBMS with mature and proven replication facilities. Does synchronising your organisatinos data hold any business value, and if so can you fight for some budget to buy the tools?

As for free stuff, I'm interested to try Open-SAP (www.sapdb.org) - it looks pretty heavy-weight for a freebie!?

If you're stuck with your own format, then you might have to write your own engine. You could also consider writing export/import routines to a commercial database and using that only for replication ( a kind of operational data store). Try looking into Oracles replication engine or some other vendors to get an idea as to how they have done replication. I'd definately research your best replication approach first though based on your circumstances, and then decide how to implement.

HTH Tobin

> Which algorithms and techniques would you recommended?
> Is there also favorable freeware software?
>
> Thanks for tipps!
> Jens
>
> fup2 comp.databases.theory
Received on Sat Aug 03 2002 - 11:21:34 CEST