Yandex Europe Ag

Aleksey Alekseevich Stankevichus

Sergey Vladimirovich Trifonov

Denis Nikolaevich Podluzhny

Andrey Vladimirovich Fomichev

A method of determining a potential anomaly of a memory device is executable at a supervisory entity computer communicatively coupled to the memory device. The method includes, over a pre-determined period of time, determining a subset of input/output (I/O) operations having been sent to the memory device for processing, applying at least one counter to determine an actual activity time of the memory device during the pre-determined period of time, applying a pre-determined model to generate an estimate of a benchmark processing time for each one of the subset of transactions, calculating a benchmark processing time for the subset of I/O operations, generating a performance parameter based on the actual activity time and the benchmark processing time, and based on an analysis of the performance parameter, determining if the potential anomaly is present in the memory device.

Method of determining potential anomaly of memory device

A method of detecting congestion in a computer system. The computer system has a task source for generating a stream of computer-implemented tasks and a plurality of processing nodes arranged in a sequence and forming a processing pipeline for processing the stream of computer-implemented tasks, each one of the plurality of processing nodes having a respective system-unique identifier (SUID). The method is executable by the computer system. The method includes receiving a task packet by a given processing node, the task packet having a task body indicative of a respective computer-implemented task, and a writable congestion-indicating field; processing the task body of the task packet; determining a presence of an input queue of more than one task packet at the given processing node; and in response to the determining the presence of the input queue, updating the writable congestion-indicating field with the SUID of the given processing node.

Processing system and method of detecting congestion in processing system

A method of determining a potential malfunction of a memory device is executable at a supervisory entity computer communicatively coupled to the memory device. The method includes, over a pre-determined period of time, determining a subset of input/output (I/O) operations having been sent to the memory device for processing, applying at least one counter to determine an actual activity time of the memory device during the pre-determined period of time, applying a pre-determined model to generate an estimate of a benchmark processing time for each one of the subset of transactions, calculating a benchmark processing time for the subset of I/O operations, generating a performance parameter based on the actual activity time and the benchmark processing time, and based on an analysis of the performance parameter, determining if the potential malfunction is present in the memory device.

A computer-implemented method for implementing a distributed computer system for processing data and a distributed computer system for processing the data are provided. The method includes receiving an indication of a redundancy algorithm to be used for generating redundancy-data based on the data. The method also includes receiving an indication of a failure-model algorithm to be used for confirming successful storage of the data to the client. The method also includes at a data storage moment in time instructing the set of storage devices to store the redundancy-data. The method also includes, at a data access moment time, receiving an indication of an access request for the data. The method also includes, at a data access moment time, applying the failure-model algorithm onto notifications received from the set of storage devices for non-ambiguously determining whether a confirmation of successful storage of the data has been provided to the client.

Method and distributed computer system for processing data

An operations scheduler and a method of scheduling I/O operations to be processed by a memory drive are disclosed. The method includes receiving I/O operations, and for each one of them determining an estimated processing time, and an estimated processing period indicative of an estimated deadline. The method also includes determining a scheduled order of I/O operations based on the respective estimated deadlines. The method also includes monitoring an actual processing time for the I/O operations and adjusting a virtual time of the operations scheduler if a given actual processing time is above a given estimated processing time of a given I/O operation.

Method and system for scheduling I/O operations for processing

A hybrid scheduler for and a method of scheduler I/O operations for execution are disclosed. The hybrid scheduler has a first a second scheduler. The method includes receiving I/O operations from a first and second I/O source and executing, by the first scheduler, a first algorithm to determine a scheduling order of the I/O operations based on the bandwidth of a memory drive. The method also includes transmitting this order of I/O operations to the second scheduler that executes a second algorithm to determine a rescheduled order of I/O operations based on respective execution deadlines. The method also includes monitoring whether execution of the I/O operations respects the respective execution deadlines. In response to determining that the execution of at least one I/O operation does not respect the respective execution deadline, the method includes pausing accepting additional I/O operations by the second scheduler from the first scheduler.

Method and system for scheduling i/o operations for execution

A method and a system for scheduling transmission of a plurality of input/output (I/O) operations to be executed by a storage device comprising: applying a slidable cell window to sequential I/O operations from a scheduled sequence, the slidable cell window being positioned over the sequential I/O operations starting with the first I/O operation, transmitting to the storage device the sequential I/O operations, receiving an operation execution confirmation from the storage device, determining a position of a given cell of the slidable cell window associated with the I/O operation, responsive to a position of the given cell being the first position of the slidable cell window: sliding the slidable cell window over the plurality of I/O operations so as to include at least one additional I/O operation in the slidable cell window, and transmitting to the storage device, the at least one additional I/O operation for execution.

Growing community of inventors

Moscow, Russia

Aleksey Alekseevich Stankevichus