Agree. Don't rely on OS to save you from fail over. Even AIX HACMP is NOT a great solution, working on this I have seen a lot of horror even in fail over test. I know of financial institution which loss $17 million in 3 hours of failover failure and eventually migrated to Tandem mainframe which has a proven 24 x 7 reliability, the same system behind the back end of NYSE.
Consider Migz and Ariz suggestion and if your company can afford, use Oracle RAC for DB. On Tue, Apr 15, 2008 at 7:36 PM, Ariz Jacinto <[EMAIL PROTECTED]> wrote: > having used those two before, the switch over may take as fast as > 6-10(?)secs but is usually set to 3 minutes to avoid false positives. what > are the chances for your user to be caught within that 10secs window? that > depends onriz suggestion the number of users at any given time :) > > my point is, the app should never assume a seamless operation of any of > its dependencies. the app should know how to fail gracefully. > > > > > On Tue, Apr 15, 2008 at 1:07 PM, <[EMAIL PROTECTED]> wrote: > > > I think this is a job for DRBD and HeartBeat > > > > [...] > > > > _________________________________________________ > Philippine Linux Users' Group (PLUG) Mailing List > http://lists.linux.org.ph/mailman/listinfo/plug > Searchable Archives: http://archives.free.net.ph > -- ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ sometimes truth is stranger than fiction -bad religion- http://www.bloglines.com/blog/mailist ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ I don't think the computers will take over the world. I have a bucket of water.
_________________________________________________ Philippine Linux Users' Group (PLUG) Mailing List http://lists.linux.org.ph/mailman/listinfo/plug Searchable Archives: http://archives.free.net.ph
