摘要:
A computer operating system that allows legacy applications to be run automatically with quality of service (QoS) guarantees matching required QoS performance levels. In accordance with the invention, files have QoS requirement attributes that can be set-by users. Additionally, users may interpose a requirement broker between a given legacy application and the operating system. The requirement broker may be in the form of a modified version of a library that is dynamically linked with applications at load time. The requirement broker intercepts certain system calls and automatically requests from the system QoS guarantees in accordance with the QoS requirement attributes of the accessed files, whether local or remote.
摘要:
A uniform application programming interface (API) provides efficient generation and control of resource reservations so as to ensure a desired quality of service (QoS) for applications running on an operating system of a computer system. The uniform API is based on a hierarchical file system which associates resource reservations with references to shared objects, rather than with the objects themselves. The uniform API permits the use of a variety of different proportional share schedulers for controlling access to physical resources, e.g., CPU, memory, disk, network interfaces, etc. of the computer system. The hierarchical file system includes a separate directory for each independently-scheduled physical resource of the computer system. A parent of the resource reservation in the hierarchical file system is either a root node of the file system or another reservation for the same resource. The resource reservation represents either an internal reservation, corresponding to a directory in the file system which is permitted to have child nodes associated therewith in the file system, or a queue, corresponding to a directory which is not permitted to have child nodes associated therewith. A given request for a resource reservation is processed by a proportional share scheduler associated with the resource, and identifies a queue having a predetermined portion of the resource allocated thereto. The invention also provides mechanisms for request tagging by the operating system, parental limitation of resources used by a child process, and “garbage collection” of resource reservations that are no longer needed.
摘要:
Received packets are processed utilizing protocol processing stages in an operating system, in accordance with a technique referred to as signaled receiver processing (SRP). A given protocol processing stage uses a next stage submit (NSS) function to pass a packet to a subsequent protocol processing stage. Each protocol processing stage is associated with one or more sockets. Each socket is associated with the input function of a protocol processing stage, an unprocessed input queue (UIQ) that contains packets destined to the socket but not yet processed by the socket's input function, and a list of owner processes, which expect input from the socket. NSS uses a multi-stage early demultiplexing (MED) function, which determines a packet's next destination socket, based on the stage currently processing the packet and on information contained in the packet. If the amount of buffering used by a packet plus the amount of input buffering already used by the packet's destination socket exceeds a designated limit associated with that socket, NSS drops the packet. Otherwise, NSS enqueues the packet in that socket's UIQ and signals that socket's owner processes. The signal triggers a default action corresponding to performance of the required protocol processing by one of the destination socket's owner processes, but those processes may override the default action by, e.g., deferring the required protocol processing until execution of a subsequent receive call.
摘要:
A method and system for servicing disk I/O requests from domains which have been guaranteed minimum quality of disk service maintains I/O requests for each domain are in separate queues. The queues are serviced by a disk scheduler which selects requests from the queues in accordance with a fair queuing scheduling algorithm that considers the estimated time required to service the request at the head of a queue with regard for the size of the input or output associated with the request, the proportion of disk bandwidth assigned to the particular domain, and the state of the other queues.
摘要:
A new scheduling method and policy for shared (server) resources, such as the CPU or disk memory of a multiprogrammed data processor. The scheduling is referred to as Move-To-Rear List Scheduling and it provides a cumulative service guarantee and well as more traditional guarantees such as fairness (proportional sharing) and bounded delay. In typical operation, a list is maintained for a server of processes seeking service from the server. Processes are admitted to the list only when maximum capacity constraints are not violated, and once on the list, are served in a front-to-back order. After receiving service, or upon the occurrence of other events, the position of the process on the list may be changed.
摘要:
A protocol and system for securely communicating a financial transaction between a customer and a merchant and a distributed computer network employing the protocol or system. A central authority having a central authority private key Ksca assigns a customer account (“CACCT”) to the customer and a merchant account (“MACCT”) to the merchant. The customer has a customer public key Kpc; the merchant has a merchant public key Km. The protocol includes the steps of: (1) sending a quotation from the merchant to the customer, the quotation including the Kpm, a Ksca-signed signature that is a function of the MACCT, an unsigned copy of a price and a Ksm-signed signature that is a function of the MACCT and the price, (2) replying to the quotation by sending an order from the customer to the merchant, the order including the Kpc, a Ksca, signed signature that is a function of the CACCT, an unsigned copy of the price and a Ksc-signed signature that is a function of the CACCT, the MACCT and the price and (3) replying to the order by filling the order.
摘要:
A computer operating system in which transfer of control of executing threads between protection domains is implemented using specific portals dedicated to the particular transfer of control between two protection domains. The transfer of control may also include parameter manipulation. A server registers a portal specification with a portal manager, the portal specification defining the behavior of portals which transfer control to the server. Thereafter, when a client application desires service from the server, the client application requests instantiation of a portal which will transfer control of an execution thread from the client application protection domain to the server protection domain. Upon receipt of the request, the portal manager dynamically generates portal code and updates a portal table associated with the client application, thus instantiating the portal. When the client application desires to invoke the server, the client application issues a system call instruction which results in switching the processor to privileged mode, followed by a lookup to the portal table and a transfer of control to the portal code identified in the portal table. The execution of the portal code results in transfer of control of the execution thread to the protection domain of the server. Various parameter manipulation techniques are disclosed for improving the efficiency of the transfer of control.
摘要:
A system for, and method of, ensuring serialization of lazy updates in a distributed database described by a directed acyclic copy graph. In one embodiment, the system includes: (1) a timestamp module that creates a unique timestamp for each of the lazy updates and (2) a propagation module, associated with the timestamp module, that employs edges of the directed acyclic copy graph to propagate the lazy updates among replicas in the distributed database according to said unique timestamp and ensure the serialization.
摘要:
An apparatus and method provide an interrupt-based add-on service for use in a service subscription system having a central service provider and at least one subscriber unit. The interrupt-based system includes a database for storing subscriber information specific to at least one subscriber, where the subscriber information includes at least one subscriber selectable condition. The system further includes a computer for interrupting a service request when the service request satisfies at least one subscriber selectable condition and for making a service connection when the service request does not match the subscriber selectable condition.
摘要:
The invention concerns a method for estimating characteristic information of data items in a data set, such as a database, based on parameters of a multifractal distribution. The invention facilitates efficient estimation of such characteristic information of data contained in a data set more accurately than known estimation methods and without requiring an exhaustive analysis of the data. The invention also concerns an efficient technique for generating the parameters for the multifractal distribution.