Re: Synchronous Replication as backup site...

Yosi

Thought I recognized your name from internal emails! I haven't played much with synchronous replication because so few clients need it, but it may be worth checking the Support site (http://support.oracle.com) or technet (http://technet.oracle.com). They had scripts for setting up and managing asynch replication, and there may be equivalent scripts for the synchronous stuff.

HTH. Pete

Yosi Greenfield wrote:

> Pete,
>
> Thanks for replying. (I left Oracle Corp about a month ago, and I miss the
> email lists desperately!) Their servers would be on a pretty solid LAN, so
> synch in and of itself should work. I've never actually seen anyone use synch
> rep though because we (Oracle) usually recommend asynch.
>
> Add to that, the goal is to backend a website with these dbs. Synch is desired
> so that two successive hits to the website will see the correct, up-to-date
> modifications made by the user, even if the hit was sent to the second server.
>
> Do you have, or could you find, scripts that automate the response to a failure
> of a node in a synchrep scenario? And are there any special gotchas for synch
> rep being the backup scheme? Any other ideas? (Did I mention I miss the email
> lists?!!)
>
> Thanks again,
>
> Yosi
>
> Pete Sharman wrote:
>
> > Yosi
> >
> > Your thinking is largely correct. What sort of comms line do they have
> > between the two sites? I would normally not recommend using synchronous
> > replication for a real time system, because they normally can't handle any
> > downtime issues. If the network goes across anything less than an internal
> > LAN, I think they need to revisit their thinking.
> >
> > HTH.
> >
> > Pete
> >
> > Yosi Greenfield wrote:
> >
> > > All,
> > >
> > > Has anyone used synchronous replication as backup / failover? Using
> > > asynch is pretty straightforward. Site goes down. You keep working on
> > > remaining site. Site comes back up, any changes are propagated.
> > >
> > > However, our user very much insists he needs real time replication.
> > > Then, if a site goes down, activity is dead. You've got to remove the
> > > down master from the group. (And in case the downed master is the master
> > > def, you've got to change master defs first. I'm thinking as I type..)
> > >
> > > Has anyone done this? If yes, do you have procedures or scripts to
> > > implement this? Any help or thoughts would be most appreciated.
> > >
> > > Thanks in advance,
> > >
> > > Yosi Greenfield
> > > yosi_at_newsalert.com
> >
> > --
> > Regards
> >
> > Pete

--
Regards

Pete

Received on Wed Jul 14 1999 - 11:22:12 CDT