Abstract: A method and apparatus are disclosed for efficiently creating and maintaining a remembered set in a generational garbage collection scheme using a write buffer and a barrier bit associated with each object. A barrier bit associated with each object differentiates generations in a generational garbage collection scheme. When an object is first created, the barrier bit of the object is set to zero. The barrier bit is set when the object becomes old, for example, after surviving a first garbage collection. The barrier bit is used to determine whether to make an entry into a write buffer when a reference to another object is stored into an object. An entry is made in the write buffer only if the barrier bit of the object that is written into is set. At the end of each garbage collection, entries in the write buffer are added to the remembered set for the next garbage collection if the objects satisfy the criterion for membership, i.e., they are live and may contain pointers to objects in a younger generation.

Abstract: A TCP-connection-router performs encapsulated clustering by dividing each encapsulated cluster into several Virtual EC (VECs), dynamically distributing incoming connections within a VEC based on current server load metrics according to a configurable policy. In one embodiment, the connection router supports dynamic configuration of the cluster, and enables transparent recovery which provides uninterrupted service to the VEC clients.

Abstract: An object, of this invention is to improve the overall throughput of an Encapsulated Cluster.

Abstract: A method and apparatus are provided for the efficient management of remembered sets in a generational garbage collection scheme. The present invention detects when an old object has a pointer to a young object, and needs to be added to the remembered set. A write buffer and a temporary buffer are used to create and maintain the remembered set. Entries in the write and temporary buffers are used as part of the root set for creating the remembered set for the next garbage collection. A barrier bit associated with each object differentiates generations in the generational garbage collection scheme and is used to determine whether to make an entry into a write buffer when a reference to another object is stored into an object. Objects that have survived one or more collections, but not the minimum number, N, of collections to be considered an old object are referred to as “middle-aged” objects. During a minor garbage collection, the write buffer is scanned.

Abstract: A method and apparatus are provided for the efficient management of remembered sets in a generational garbage collection scheme. In order to manage the remembered set, the present invention provides a first mechanism for detecting when an old object has a pointer to a young object, and needs to be added to the remembered set, and a second mechanism for detecting when an object already in the remembered set no longer contains a pointer to a young object, so that the object can be removed from the remembered set. Entries in the remembered set have an associated counter, C, identifying the garbage collection cycle during which the object was placed in the remembered set. Objects inserted into the remembered set are assumed to point to the youngest possible object. Entries automatically expire from the remembered set when the garbage collection counter reaches C+N, since the objects pointed to by the object in the remembered set must now themselves be old objects.

Abstract: The invention provides a method for compacting transactions which have committed at a disconnected client prior to sending the transactions to the server for replaying along with an iterative process for re-calculating the combined transaction in the event that replay at the server fails.

Abstract: A TCP-connection-router performs encapsulated clustering by dividing each encapsulated cluster into several Virtual EC (VECs), dynamically distributing incoming connections within a VEC based on current server load metrics according to a configurable policy. In one embodiment, the connection router supports dynamic configuration of the cluster, and enables transparent recovery which provides uninterrupted service to the VEC clients.

Abstract: A system and method for recovering from failures in the disk access path of a clustered computing system. Each node of the clustered computing system is provided with proxy software for handling physical disk access requests from applications executing on the node and for directing the disk access requests to an appropriate server to which the disk is physically attached. The proxy software on each node maintains state information for all pending requests originating from that node. In response to detection of a failure along the disk access path, the proxy software on all of the nodes directs all further requests for disk access to a secondary node physically attached to the same disk.