The abrupt failure of the server machine running the database is the most common cause of unplanned downtime. The server may crash because of hardware problems, such as the failure of a power supply, or because of software problems, such as a process that begins to consume all the machine’s CPU resources. Even if the underlying server platform is fine, the Oracle instance itself can fail. Whatever the cause of the crash, the effect on Oracle is the same—the instance ceases to exist. Note that it’s the instance that crashes, not the database. As you will recall from Chapter 2, the Oracle instance runs on the server and provides access to the database, which is stored on disk. This means that even a system crash, by itself, will not imperil any data that’s already safely stored within the disk files used by the Oracle database.
The impact of the crash will depend on the activity in progress at the time of the crash. Any connected sessions will no longer have a server process to which to talk. All active queries and transactions will be abruptly terminated. The process of cleaning up the resulting mess is called instance recovery or crash recovery .