Abstract: Multi-pass parallel merging in a database includes identifying characteristics of non-final pages during database query operations. A phase of page consolidation is triggered based on the identified characteristics and a final page is stored.

Abstract: Disclosed herein are system, method, and computer program product embodiments for storing and accessing data in a shared disk database system using a timestamp range to improve cache efficiency. An embodiment operates by retrieving, by a node, from a shared storage. a blockmap identity and a root page associated with a data request, based on a determination that the blockmap identity associated with a data request is present in a cache. The embodiment continues, retrieving, by the node, the logical page by copying a stored logical page from the shared storage and setting a lower timestamp value of the logical page to a timestamp associated with the stored logical page and an upper timestamp value of the logical page to a timestamp associated with the data request, based on a determination that the logical page is not present in the cache.

Abstract: Multi-pass parallel merging in a database includes identifying characteristics of non-final pages during database query operations. A phase of page consolidation is triggered based on the identified characteristics and a final page is stored.

Abstract: A distributed database system providing data and space management methodology is described. In one embodiment, for example, a method is described for detection of deadlocks in a distributed database system comprising a plurality of servers sharing access to data, the method comprises steps of: in response to a task at a first server waiting to obtain a lock regulating access to data, sending a search request including lock information to a second server holding the lock; at the second server, examining dependencies of the task holding the lock based on the lock information received with the search request and local lock information; if the task holding the lock is waiting to obtain a lock held by an additional server, forwarding the search request and local lock information to the additional server; repeating the examining and forwarding steps until all dependencies are examined; and responding to the search request indicating whether a deadlock was detected and providing lock information found.