There are two databases DB1,DB2 situated at two different placessay pl1,pl2.people at place pl1 will access DB1 and for people at pl2 access DB2.Basically these two databases same date and that is acheived currently through (perl)scripts(of the web interface which write to both the databases at a time).It is bugged when one server is down no insert or update will work(according to scripts).
Now i have to make this system more redundent by finding a solution through oracle(involved as a DBA ).I need suggesion in how to tackle the situation such that give them more redundent solution...like how shall i proceed with....will parallel server (on distributed Db systems as they r physically located at different location) or will replication would be the solution in any case the DOWNTIME MUST BE ZERO AND IN THE DUE COURSE OF BRINGING UP THE FAILED SERVER STILL THE USERS SHOULD ACCESS THE DATABASE(THERE ARE 2 MILLION USERS).(am on sun OS)
quite urgent matter and looking for the help.and also ne one can explain how clustering in oralce occurs.
Looking at your conditions, oracle parallel server would be a good choice. In replication, you have to enforce the sync to keep the data in sync. On parallel server things are done automatically. In the 9i this type of operation is called clustering.
hi Jeff Hunter and sam,
thanx for quick reply.......am on oracle 8i..... for more redundency we have two servers at each place....and we r planning to have two storages at each place and and two servers for each storage for more redundency .
There is problem with parallel server that to maintain network traffic and to maintain bandwidth..heavy traffic is expected like 2 million users r there and our RAS(remote Acces Servers and give access to 30 ,000 users at a time...basically it's like authenticating database (for userid and pwd).
though they contain same data they r two DB's and as i mentioned people at restective places will be accessing respective servers and in case of failure...all 'll be accessing one DB(we did this radius server with virual ip resolution).
so i guess we cannot afford to maintain badwidth and network traffic in can of parallel server.
Regarding replication....in case of one of failure of one other will be takking all the requests and the failed one will be up and allowed to do server only when it has the same data as that of original ..
* will replication takes care this situation in fail over situation?
* will that be assured ZERO downtime in this mean while
If so how to proceed????.......ne one help me out soon..a bit urgent ...
You want to setup a RADIUS server that authenticates against more than one Oracle server. When one site goes down, you want the users to automatically authenticate over to the second geo location. I architected this exact same setup for an ISP in the Southeastern US.
First, your RAS units probably have the ability to authenticate against a primary and secondary RADIUS Server. If not, you may want to look at a product called Steel Belted Radius. In this particular ISP's case, I specified two distinct geographic locations each with a 2 node OPS cluster. Each OPS cluster then used replication to sync it's changes with the other site on a 5 minute basis. Each RAS unit would then have the "closest" RADIUS server as the primary and the "farther" RADIUS server as the secondary.
yeah hear me to in the same situation...we do have radius servers and all that u mentioned......from ur answer this phrase
"Each OPS cluster then used replication to sync it's changes with the other site on a 5 minute basis. "
could u like to elaborate a bit at this point like is that possible to replicate synchronously....i mean this situation look s like (multi master) synchronous replication.....could u please tel how and what kind of replication is best fit in this situation and also NE HOW RADIUS SERVER SWITCHES BETWEEN IN MEAN WHILE IF ONE OF THE SERVER IS DOWN AND IN THAT TIME THE SECOND SET AT OTHER PLACE WILL TAKE CARE."IS THAT POSSIBLE THAT WHEN WE UP THE SERVER(WHICH WAS DOWN)AND THAT INSTANCE OF TIME BOTH THE SERVER MUST HAVE SAME DATA AT NE COST...(NE WAYS DOWN TIME MUST BE ZERO).....
how will this help in case of one database fails with ne reason and we bring it up....could u please explain this scenario.......