High availability config question

We are about to design a configuration of three servers and a separate array of disks, as shown in the ASCII-drawing below (best viewed with a Fixed Width font): +-----------------+ | Array of disks | | with DbSpaces | |of DB1, DB2, DB3| +-----------------+ | | | +--------------+ | +-------------+ | | | +-------------+ +-------------+ +-------------+ | Server 1 | | Server 2 | | Server 3 | | | | | | | | Informix | | Informix | | Informix | | Online | | Online | | Online | | | | | | | | Application | | Application | | Application | |connected to | |connected to | |connected to | | DB1 | | DB2 | | DB3 | +-------------+ +-------------+ +-------------+ | | | | | | | | | WK WK WK WK WK WK WK WK WK [the "Application" is the same on each server; it connects to a database named by a configuration variable, set to DB<n> on server <n>] [the "Informix Online" is installed with the same version on all servers] [WK stands for workstation, as you might have guessed] In a normal case, i.e. all machines and disks alive, each server application accesses the database defined/declared in the local Informix Online engine. The array of disks contains all dbspaces used by that configuration, possibly implementing table fragmentation strategies, for performance. My question concerns the feasibility of the following situation: If one server (NOT its disks) fails, or is stopped for maintainance, the application that was running on it should be started as a new instance on one of the remaining servers, and still access the database it worked with. For example, if the server 3 fails, the server 2 must start the application as a second instance, and access the database "DB3". It is assumed that a software/hardware mechanism has re-routed the disks connections to the remaining alive servers. Maybe a combination of such mechanism and a specific version of Informix server can help. Any ideas? I will be pleased to hear of your suggestions! Philippe Job Siemens Nixdorf Namur (Belgium) Mail: Philippe.Job@sni.be