33 results about "Asynchronous I/O" patented technology

Some embodiments concern a kernel bus system for building at least one virtual machine monitor. The kernel bus system is based on kernel-based virtual machine. The kernel bus system is configured to run on a host computer. The host computer comprising one or more processors, one or more hardware devices, and memory. The kernel bus system can include: (a) a hyperbus; (b) one or more user space components; (c) one or more guest space components configured to interact with the one or more user space components via the hyperbus; (d) one or more VMM components having one or more frontend devices configure to perform I / O operations with the one or more hardware devices of the host computer using a zero-copy method or non-pass-thru method; (e) one or more para-virtualization components having (1) a virtual interrupt configured to use one or more processor instructions to swap the one or more processors of the host computer between a kernel space and a guest space; and (2) a virtual I / O driver configured to enable synchronous I / O signaling, asynchronous I / O signaling and payload delivery, and pass-through delivery independent an QEMU emulation; and (f) one or more KVM components. The hyperbus, the one or more user space components, the one or more guest space components, the one or more VMM components, the one or more para-virtualization components, and the one or more KVM components are configured to run on the one or more processors of the host computer. Other embodiments are disclosed.

Some embodiments concern a kernel bus system for building at least one virtual machine monitor. The kernel bus system can include: (a) a hyperbus; (b) user space components; (c) guest space components configured to interact with the user space components via the hyperbus; (d) VMM components having frontend devices configure to perform I / O operations with the hardware devices of the host computer using a zero-copy method or non-pass-thru method; (e) para-virtualization components having (1) a virtual interrupt module configured to use processor instructions to swap the processors of the host computer between a kernel space and a guest space; and (2) a virtual I / O driver configured to enable synchronous I / O signaling, asynchronous I / O signaling and payload delivery, and pass-through delivery independent an QEMU emulation; and (f) KVM components. Other embodiments are disclosed.

Resource acquisition requests for a filesystem are executed under user configurable metering. Initially, a system administrator sets a ratio of N:M for executing N read requests for M write requests. As resource acquisition requests are received by a filesystem server, the resource acquisition requests are sorted into queues, e.g., where read and write requests have at least one queue for each type, plus a separate queue for metadata requests as they are executed ahead of any waiting read or write request. The filesystem server controls execution of the filesystem resource acquisition requests to maintain the ratio set by the system administrator.

Resource acquisition requests for a filesystem are executed under user configurable metering. Initially, a system administrator sets a ratio of N:M for executing N read requests for M write requests. As resource acquisition requests are received by a filesystem server, the resource acquisition requests are sorted into queues, e.g., where read and write requests have at least one queue for each type, plus a separate queue for metadata requests as they are executed ahead of any waiting read or write request. The filesystem server controls execution of the filesystem resource acquisition requests to maintain the ratio set by the system administrator.

As an I / O method of issuing an I / O request, there are a synchronous I / O method, in which a task is in a waiting state until an I / O response to the I / O request is returned and an asynchronous I / O method, in which another task can be executed before an I / O response to the I / O request is returned. When issuing an I / O request in execution of a task, a DBMS selects any one of the synchronous I / O method and the asynchronous I / O method, based on an I / O response speed of an I / O destination storage device and issues an I / O request according to the selected I / O method.

Embodiments of the present invention address deficiencies of the art in respect to threading model switching between asynchronous I / O and synchronous I / O models and provide a novel and non-obvious method, system and computer program product for autonomic threading model switching based upon I / O request types. In one embodiment, a method for autonomic threading model switching based upon I / O request types can be provided. The method can include selectably activating and de-activating a blocking I / O threading model according to a volume of received and completed blocking I / O requests.

The invention belongs to network multimedia field, specifically relates to a special disc read-write system with time-variable IPTV live server and a time-variable IPTV live server using the same, wherein the disc read-write system comprises: a media data buffer server including a channel management module, a buffer management module with memory based on tmpfs and memory file mapping, and a file read / write module with DIRECT I / O and asynchronous I / O technology; and some buffer clients including the direct data write / read client. The time-variable IPTV live server also comprises a data pre-process module, a transmission module and user management module, wherein the special disc read-write system transmits the media data from the data pre-process module to the transmission module and finally sends to every client player, so as to improve entire system performance of the time-variable IPTV live server.

Methods and systems for pre-fetching operations include executing event callbacks in an event loop using a processor until execution stops on a polling request. A path walk is performed on future events in the event loop until the polling request returns to pre-fetch information for the future events into a processor cache associated with the processor. Execution of the event callbacks in the event loop is resumed after the polling request returns.

The invention discloses an asynchronous input and output signal processing method, provides a set of event-based signal processing mechanism, and makes a special design on the interface of the driver program to cooperate with this set of signal processing, including the following steps: Step 1, as a driver program The user thread of the user puts forward the need for input and output signal processing through the driver interface; step 2, the driver creates an event object and exports it to the user thread; step 3, the user thread obtains the input and output processing signal through the event object, And complete the corresponding operation according to the definition of the signal by the driver; the driver completes the input and output operations of the device, and notifies the user thread to process the signal through the event object. Simplify some advanced synchronization mechanisms and provide them to driver designers; a user thread can perform asynchronous I / O operations with multiple devices at the same time; drivers can customize the signal semantics of asynchronous I / O to provide richer functions .

The invention discloses an I / O interface circuit of an asynchronous SRAM, belonging to the technical field of SRAM data reading and writing. The interface circuit includes a read circuit and a write circuit; the output end of the read circuit is connected to the input end of the write circuit; the read circuit is connected to the enable signal, the data signal and the read signal; the write circuit is connected to the write signal . The I / O interface circuit of the asynchronous SRAM of the present invention controls the operation of the output data by using the enable signal and the read signal, and uses the write signal to control the write operation of the data, so that the read or write of the data is realized conveniently and quickly Operation; by introducing positive feedback in the read circuit, the accuracy of the level signal is enhanced; there is no path from the power supply VDD to the ground GND, which greatly reduces the power consumption of the read circuit; by using the read signal to control the transistor Turn on to speed up the response speed of the circuit.

The system and method for distributing data in the big data platform of the present invention adopts asynchronous I / O as the technical basis to build a big data distribution unit for high-speed distribution of big data, adopts the separation of server and client threads to improve the throughput of data distribution, The complete structure of the data is guaranteed by the multi-dimensional structure storage unit, and the accuracy and correctness of the data distribution are guaranteed by the big data management center and the data bus module, so that all parts can cooperate at high speed without waiting for each other for resources, and can fully utilize resource. At the same time, the whole system shows good scalability.

The present invention provides an optimization method for compressing and uploading self-service terminal logs. The method includes: the client uses entropy coding as the basis to compress the logs to be uploaded by means of displacement operations; Some of the log files are uploaded to the server in sequence. The server receives the compressed log file uploaded by each client; the server allocates the compressed log file received to the file storage queue, and returns a normal reception response to the client; the server detects the compressed Whether there is any abnormal information in the log file; if not, then the server writes the compressed log file in the file saving queue into the disk of the server by means of asynchronous I / O. The above method effectively reduces the time for the client to wait for the response from the server, and improves the functional experience of the self-service terminal.