Patent application number | Description | Published |
20090271662 | STEADY STATE COMPUTER TESTING - A steady state value of each parameter in a set of one or more parameters of a computer system under test is defined, and acts related to the parameter are performed while the system under test is processing one or more test workloads. The acts can include comparing a working value of the parameter with the steady state value of the parameter, as well as operating on one or more workloads to bring the working value closer to a steady state value. This can include increasing or decreasing the working value of the parameter, such as by admitting one or more workloads to the system under test or cancelling one or more workloads from the system under test. | 10-29-2009 |
20090307530 | DISTRIBUTED TESTING SYSTEM AND TECHNIQUES - A fully distributed system for testing a distributed computing system utilizes test services exposed in conjunction by independent test components deployed on one or more computing nodes of the distributed computing system. The test services ensure dynamic and adaptive coordination amongst the test components. The test services include but are not limited to: an atomic synchronization service; a distributed shared memory service; a lock manager service; a publish/subscribe service; and a producer/consumer service. | 12-10-2009 |
20120101800 | MODEL CHECKING FOR DISTRIBUTED APPLICATION VALIDATION - A model checking system is described herein that more effectively verifies and validates the design of distributed applications by providing a model and a generic framework to check application invariant properties, detect anomaly behaviors, and monitor application health. The model checking system checks on-line application behavior against application models derived from formal descriptions of the application. The system formulates the concrete application as an abstract model and a number of rules or properties that are expected to hold for the application under all conditions. The model checker compares the actual application execution with the models and either confirms that the properties hold true or reports that the properties are violated. Thus, the model checking system provides more efficient and thorough validation of distributed applications under more realistic production conditions. | 04-26-2012 |
20140108483 | DISTRIBUTED APPLICATION OPTIMIZATION USING SERVICE GROUPS - Embodiments are directed to managing multiple different types of applications using service groups. In one scenario, a computer system receives an indication of one or more application dependencies and characteristics that are to be implemented when an application is provisioned on a distributed host computer system. The computer system creates an application manifest that declaratively defines application dependencies and characteristics for various different service groups. Each service group includes applications that match the declaratively defined application dependencies and characteristics. The computer system also sends the manifest to the distributed host computer system which loads those applications that fit the manifest criteria onto available nodes of the distributed host computer system according to the service groups specified in the manifest. | 04-17-2014 |
Patent application number | Description | Published |
20110296069 | Fabric Based Lock Manager Service - A replicated finite state machine lock service facilitates resource sharing in a distributed system. A lock request from a client identifies a resource and a lock-mode, and requests a leaseless lock on the resource. The service uses client instance identifiers to categorize requests as duplicate, stale, abandoned, or actionable. A lock may be abandoned when a client holding the lock goes down. After a per-client abandonment timer expires, the lock service may treat any exclusive lock granted to the client as abandoned, and treat any non-exclusive lock granted to the client as unlocked. The service tries to notify a lock-holding client if another client requests the same lock, and treats the lock as abandoned if the notification attempt fails. An abandoned read lock is granted to a different client on request. An abandoned write lock is granted or refused depending on whether the requesting client accepts abandoned write locks. | 12-01-2011 |
20120079234 | PERFORMING COMPUTATIONS IN A DISTRIBUTED INFRASTRUCTURE - The present invention extends to methods, systems, and computer program products for performing computations in a distributed infrastructure. Embodiments of the invention include a general purpose distributed computation infrastructure that can be used to perform efficient (in-memory), scalable, failure-resilient, atomic, flow-controlled, long-running state-less and state-full distributed computations. Guarantees provided by a distributed computation infrastructure can build upon existent guarantees of an underlying distributed fabric in order to hide the complexities of fault-tolerance, enable large scale highly available processing, allow for efficient resource utilization, and facilitate generic development of stateful and stateless computations. A distributed computation infrastructure can also provide a substrate on which existent distributed computation models can be enhanced to become failure-resilient. | 03-29-2012 |
20120079505 | PERFORMING COMPUTATIONS IN A DISTRIBUTED INFRASTRUCTURE - The present invention extends to methods, systems, and computer program products for performing computations in a distributed infrastructure. Embodiments of the invention include a general purpose distributed computation infrastructure that can be used to perform efficient (in-memory), scalable, failure-resilient, atomic, flow-controlled, long-running state-less and state-full distributed computations. Guarantees provided by a distributed computation infrastructure can build upon existent guarantees of an underlying distributed fabric in order to hide the complexities of fault-tolerance, enable large scale highly available processing, allow for efficient resource utilization, and facilitate generic development of stateful and stateless computations. A distributed computation infrastructure can also provide a substrate on which existent distributed computation models can be enhanced to become failure-resilient. | 03-29-2012 |
20130290398 | PERFORMING COMPUTATIONS IN A DISTRIBUTED INFRASTRUCTURE - The present invention extends to methods, systems, and computer program products for performing computations in a distributed infrastructure. Embodiments of the invention include a general purpose distributed computation infrastructure that can be used to perform efficient (in-memory), scalable, failure-resilient, atomic, flow-controlled, long-running state-less and state-full distributed computations. Guarantees provided by a distributed computation infrastructure can build upon existent guarantees of an underlying distributed fabric in order to hide the complexities of fault-tolerance, enable large scale highly available processing, allow for efficient resource utilization, and facilitate generic development of stateful and stateless computations. A distributed computation infrastructure can also provide a substrate on which existent distributed computation models can be enhanced to become failure-resilient. | 10-31-2013 |
20140095930 | ENHANCED ARBITRATION PROTOCOL FOR NODES IN A CLUSTER - Arbitrating node failures. A method includes determining at a first node that communication to a second node has been lost. The method further includes sending an arbitration request to a plurality of arbitrators. The method further includes receiving from the plurality of arbitrators a sufficient number of arbitration reply grant messages to have arbitration reply grant messages from a quorum of the arbitrators. As a result of receiving a sufficient number of arbitration reply grant messages to have arbitration reply grant messages from a quorum of the arbitrators, the method further includes declaring the second node down. | 04-03-2014 |
20140208173 | COMMON LEASE AGENT FOR CLUSTER COMMUNICATION - Detecting failures of interconnected machines. A method includes establishing a machine to machine lease at a first machine using a first lease agent at the first machine to a second lease agent at a second machine as a result of an application on the first machine requesting to establish a lease with an application on the second machine. Using the machine to machine lease, the method further includes detecting a communication failure between the first machine and the second machine or a machine failure of the second machine. | 07-24-2014 |