Fault tolerance thru redundancy

[psganesh]

Hi,

I wish to maintain an identical copy of my database in a seperate server (ie, all inserts & updates will be done in both the databases simultaneously for each transaction)

this is to ensure availability of data in case anyone server fails.

the app has to run 24x7.

now i want your help for the following

1.how to identify if a server has failed (may be h/w failure)

2.upon identification, the app must send alert-email to the admin and after the identification, all modification of data must be done in the server that is ok.

3.Once the failed server is set right (say, harddisk is replaced), the admin will inform the application that the server is ok, and the app will set the server updated with all the data.

pls help me as to how to go about this, so as to ensure 24x7 availability of the application, even if one of the servers fails.


thankx and regards,
psganesh

[varg]

Hi psganesh,

From your requirements, I'd say you would want to look at shared-nothing clustering. With this solution implemented at the platform level, you can then identify if a particular server/node has failed (and with tivoli, Big Brother or one of the many other system monitoring tools) be able to identify if any specific resource has failed.

So, points 1-2 are covered with the monitoring solution you choose (many available), and point 3 is taken care of by the clustering solution.

Hope this helps a little

Cheers
Rob