摘要:
A method for managing packet traffic in a data processing network includes collecting data indicative of the amount of packet traffic traversing each of the links in the network's interconnect. The collected data includes source and destination information indicative of the source and destination of corresponding packets. A heavily used links are then identified from the collected data. Packet data associated with the heavily used link is then analyzed to identify a packet source and packet destination combination that is a significant contributor to the packet traffic on the heavily used link. In response, a process associated with the identified packet source and packet destination combination is migrated, such as to another node of the network, to reduce the traffic on the heavily used link. In one embodiment, an agent installed on each interconnect switch collects the packet data for interconnect links connected to the switch.
摘要:
In one embodiment, a control system supports an unlimited number of feedback control loops all sharing control of a component. A component performance rate or “speed” is used as a common metric for negotiating control of the component. Each control loop continuously monitors a system parameter it is tasked with regulating, compares it to a setpoint for that system parameter, and “requests” a speed in relation to the deviation of the associated system parameter from the corresponding setpoint. A controller receives the requested speeds as dynamic inputs and selects one of the requested speeds according to predefined selection logic. The controller communicates the selected speed to an actuator, which causes the component to operate at the selected speed. In this manner, the control system in effect negotiates control of the component in a way that ensures that all of the system parameters are being managed within safe limits.
摘要:
A data processing system includes a power supply, a plurality of processors wherein each processor is separately powerable by the power supply under operating system control. The operating system determines periodically a measure of system utilization and controls the switches to alter the number of active (powered) processors where the number of active processors reflects the measured system utilization and a set of utilization threshold values. System utilization may be based on the number of active tasks. The utilization thresholds preferably include a maximum threshold and a minimum threshold. A measured utilization exceeding the maximum threshold causes an increase in the number of active processors while utilization less than the minimum threshold causes a decrease in the number of active processors. The utilization thresholds may be determined from threshold factors that reflect time and date information, quality of service information, or a weighted average of historical utilization values.
摘要:
A data processing system includes a power supply, a plurality of processors wherein each processor is separately powerable by the power supply under operating system control. The operating system determines periodically a measure of system utilization and controls the switches to alter the number of active (powered) processors where the number of active processors reflects the measured system utilization and a set of utilization threshold values. System utilization may be based on the number of active tasks. The utilization thresholds preferably include a maximum threshold and a minimum threshold. A measured utilization exceeding the maximum threshold causes an increase in the number of active processors while utilization less than the minimum threshold causes a decrease in the number of active processors. The utilization thresholds may be determined from threshold factors that reflect time and date information, quality of service information, or a weighted average of historical utilization values.
摘要:
In one embodiment, a control system supports an unlimited number of feedback control loops all sharing control of a component. A component performance rate or “speed” is used as a common metric for negotiating control of the component. Each control loop continuously monitors a system parameter it is tasked with regulating, compares it to a setpoint for that system parameter, and “requests” a speed in relation to the deviation of the associated system parameter from the corresponding setpoint. A controller receives the requested speeds as dynamic inputs and selects one of the requested speeds according to predefined selection logic. The controller communicates the selected speed to an actuator, which causes the component to operate at the selected speed. In this manner, the control system in effect negotiates control of the component in a way that ensures that all of the system parameters are being managed within safe limits.
摘要:
A method for managing power in a data processing system having multiple components includes determining a power budget for the system. Activity levels during a forthcoming time interval are then predicted for each of the components. Using the predicted activity levels, the power budget is allocated among the system components. An activity limit is then established for each component based on its corresponding portion of the power budget. The activity of a component is then monitored and, if the component's activity exceeds the component's corresponding activity limit, constrained. Determining the predicted level of activity may include determining a predicted number of instructions dispatched by a processor component or a predicted number of memory requests serviced for a system memory component. Allocating the power budget includes allocating each component its corresponding standby power and a share of the system power available for dynamic powering based on the expected levels of activity.
摘要:
A method, system, and computer program product for compacting a non-biased results multiset are provided in the illustrative embodiments. A set of references and a multiset of values are identified. The multiset includes a first and a second set of values, each set including a first value. A first reference in the set of references refers to the first set of values and a second reference in the set of references refers to the second set of values. The values in the first and second set of values are re-arranged to form permuted first and second sets of values. The multiset is compacted by overlaying the permuted first and second sets of values in a portion such that the permuted first set of values and the permuted second set of values share a single instance of the first value in a portion of the compacted multiset.
摘要:
A method for managing packet traffic in a data processing network includes collecting data indicative of the amount of packet traffic traversing each of the links in the network's interconnect. The collected data includes source and destination information indicative of the source and destination of corresponding packets. A heavily used links are then identified from the collected data. Packet data associated with the heavily used link is then analyzed to identify a packet source and packet destination combination that is a significant contributor to the packet traffic on the heavily used link. In response, a process associated with the identified packet source and packet destination combination is migrated, such as to another node of the network, to reduce the traffic on the heavily used link. In one embodiment, an agent installed on each interconnect switch collects the packet data for interconnect links connected to the switch.
摘要:
A method for managing power in a data processing system having multiple components includes determining a power budget for the system. Activity levels during a forthcoming time interval are then predicted for each of the components. Using the predicted activity levels, the power budget is allocated among the system components. An activity limit is then established for each component based on its corresponding portion of the power budget. The activity of a component is then monitored and, if the component's activity exceeds the component's corresponding activity limit, constrained. Determining the predicted level of activity may include determining a predicted number of instructions dispatched by a processor component or a predicted number of memory requests serviced for a system memory component. Allocating the power budget includes allocating each component its corresponding standby power and a share of the system power available for dynamic powering based on the expected levels of activity.
摘要:
A method for managing power in a data processing system having multiple components includes determining a power budget for the system. Activity levels during a forthcoming time interval are then predicted for each of the components. Using the predicted activity levels, the power budget is allocated among the system components. An activity limit is then established for each component based on its corresponding portion of the power budget. The activity of a component is then monitored and, if the component's activity exceeds the component's corresponding activity limit, constrained. Determining the predicted level of activity may include determining a predicted number of instructions dispatched by a processor component or a predicted number of memory requests serviced for a system memory component. Allocating the power budget includes allocating each component its corresponding standby power and a share of the system power available for dynamic powering based on the expected levels of activity.