A method of distributing power among servers is described includes calculating thermal multipliers of the servers, where the thermal multipliers represent recommended thermodynamic power consumption levels for the servers. In addition, the thermal multipliers of the servers are discretized to one of
A method of distributing power among servers is described includes calculating thermal multipliers of the servers, where the thermal multipliers represent recommended thermodynamic power consumption levels for the servers. In addition, the thermal multipliers of the servers are discretized to one of a plurality of server power states based upon geographic locations of the servers and one of the plurality of server power states is assigned to the servers based upon the discretization of the thermal multipliers.
대표청구항▼
What is claimed is: 1. A method of distributing power among servers, said method comprising: calculating thermal multipliers of the servers, wherein the thermal multipliers represent recommended thermodynamic power consumption levels for the servers; discretizing the thermal multipliers of the serv
What is claimed is: 1. A method of distributing power among servers, said method comprising: calculating thermal multipliers of the servers, wherein the thermal multipliers represent recommended thermodynamic power consumption levels for the servers; discretizing the thermal multipliers of the servers to one of a plurality of server power states based upon geographic locations of the servers; and assigning one of the plurality of server power states to the servers based upon the discretization of the thermal multipliers; and wherein said step of discretizing the thermal multipliers of the servers comprises determining servers having thermal multipliers close to a first power state; and for a first poacher server, poaching power from at least one first poachee server having a thermal multiplier below the first power state. 2. The method according to claim 1, further comprising: receiving outlet temperatures of the servers; receiving at least one CRAC unit supply air temperature; and wherein the step of calculating the thermal multipliers further comprises computing the thermal multipliers based upon a ratio of the at least one CRAC unit supply air temperature and the outlet temperatures of the servers. 3. The method according to claim 2, wherein the step of calculating the thermal multipliers (Pi) for each of the servers (i) is based upon at least one of thermodynamic modeling, dimensionless parameters, re-circulation, and history-based profiling. 4. The method according to claim 1, wherein the step of calculating the thermal multipliers further comprises calculating the thermal multipliers during an operation time using monitoring. 5. The method according to claim 1, wherein the step of calculating the thermal multipliers further comprises calculating the thermal multipliers through modeling. 6. The method according to claim 1, wherein the at least one first poachee server is in relative close proximity to the first poacher server. 7. The method according to claim 6, further comprising: updating the thermal multiplier of the at least one poachee server by a first poaching factor, wherein the first poaching factor is a function of a distance between the first poacher server and the at least one first poachee server; and marking the at least one first poachee server as having been poached. 8. The method according to claim 1, wherein the step of discretizing the thermal multipliers of the servers further comprises an iterative multi-step procedure, said iterative multi-step procedure further comprising: discretizing the thermal multipliers of the servers based upon geographically clustered collections of racks in increasing granularity. 9. The method according to claim 1, wherein the step of discretizing the thermal multipliers of the servers further comprises an iterative multi-step procedure, said iterative multi-step procedure further comprising: discretizing the thermal multipliers based upon a sequential ordering of the thermal multipliers. 10. The method according to claim 9, wherein the step of discretizing the thermal multipliers based upon a sequential ordering of the thermal multipliers further comprises using the highest thermal multiplier first in discretizing the thermal multipliers. 11. The method according to claim 1, wherein the step of discretizing the thermal multipliers of the servers further comprises an iterative multi-step procedure, said iterative multi-step procedure further comprising: calculating the target power consumption levels for a plurality of racks in a row of racks; and determining the one of the plurality of server power states to assign to the servers contained in the plurality of racks. 12. The method according to claim 11, wherein the step of determining the one of the plurality of power states to assign comprises: calculating thermal multipliers for the servers based upon the calculated target power consumption levels for a plurality of racks; creating power budgets for the servers contained in the plurality of racks; and determining whether the power budget for a first server in a first rack of the plurality of racks exceeds at least one of the plurality of server power states. 13. The method according to claim 12, further comprising: in response to the power budget for the first server exceeding a first server power state of the plurality of server power states, assigning the first server the first server power state; and adjusting a cumulative power budget for a second rack based upon a difference between the cumulative power budget for the first rack and the first server power state. 14. The method according to claim 12, further comprising: in response to the power budget for the first server falling below the plurality of server power states, adding the power budget for the first server to the power budget for a second servers and determining whether the cumulative power budget for the first server and the second server exceeds at least one of the plurality of plurality of server power states. 15. The method according to claim 14, further comprising: in response to the cumulative power budget exceeding a first server power state of the plurality of server power states, assigning the first server a zero power level; assigning the second rack the first server power state; and adjusting the cumulative power budget for a third server based upon a difference between the cumulative power budget and the first server power state. 16. The method according to claim 11, wherein the at least one first poachee server having a thermal multiplier below the first power state is in relative close proximity to the first poacher server, and wherein the step of determining the one of the plurality of power states to assign the servers contained in the plurality of racks further comprises: determining servers having thermal multipliers below a second power state, wherein the second power state is higher than the first power state; for a second poacher server, poaching power from at least one second poachee server having a thermal multiplier below the second power state, wherein the at least one second poachee server having thermal multiplier below the second power state is in relative close proximity to the second poacher server; updating the thermal multiplier of the at least one second poachee server having a thermal multiplier below the second power state by a second poaching factor, wherein the second poaching factor is a function of a distance between the second poacher server and the at least one second poachee server having a thermal multiplier below the second power state; and marking the at least one second poacher server having a thermal multiplier below the second power state as having been poached. 17. A method of distributing power among servers, said method comprising; calculating thermal multipliers of the servers, wherein the thermal multipliers represent recommended thermodynamic power consumption levels for the servers; discretizing the thermal multipliers of the servers to one of a plurality of server power states based upon geographic locations of the servers; assigning one of the plurality of server power states to the servers based upon the discretization of the thermal multipliers; wherein the step of discretizing the thermal multipliers of the servers comprises an iterative multi-step procedure, said iterative multi-step procedure further comprising; calculating the target power consumption levels for a plurality of racks in a row of racks; and determining the one of the plurality of power states to assign to the servers contained in the plurality of racks; wherein the step of calculating the target power consumption levels comprises: calculating thermal multipliers for the plurality of racks; creating a cumulative power budget for the plurality of racks, wherein the cumulative power budget for a rack in the plurality of racks comprises a sum of the thermal multipliers of the servers contained in the rack; and determining whether the cumulative power budget for a first rack in the plurality of racks exceeds an assignable threshold power level. 18. The method according to claim 17, further comprising: assigning the first rack the assignable threshold power level in response to the cumulative power budget for the first rack exceeding the assignable threshold power level; and adjusting the cumulative power budget for a second rack based upon a difference between the cumulative power budget for the first rack and the assignable threshold power level. 19. The method according to claim 17, further comprising: adding the cumulative power budget for the first rack to the cumulative power budget for a second rack in response to the cumulative power budget for the first rack falling below the assignable threshold power level; and determining whether the added cumulative power budget for the second rack exceeds the assignable threshold power level. 20. A system for distributing power among servers, said system comprising: server temperature sensors configured to detect the temperatures of airflow exhausted from the servers; a CRAC unit temperature sensor configured detect the temperature of airflow supplied by the CRAC unit; a resource manager configured to receive the detected server exhaust temperatures and the CRAC unit supply temperature and to calculate thermal multipliers for the servers based upon the respective detected server exhaust temperatures and the CRAC unit supply air temperature, said resource manager being further configured to discretize the thermal multipliers of the servers to one of a plurality of server power states based upon geographic locations of the servers, and wherein the resource manager is configured to distribute power by assigning one of the plurality of server power states to the servers based upon the discretization of the thermal multipliers; wherein the resource manager is further configured to determine servers having thermal multipliers below a first power state and to poach power from at least one first poachee server having a thermal multiplier below the first power state. 21. The system according to claim 20, wherein the resource manager is further configured to calculate the thermal multipliers (Pi) for each of the serves (i) based upon at least one of thermodynamic modeling dimensionless parameters, re-circulation, and histoy-based profiling. 22. The system according to claim 20, wherein the resource manager is further configured to give the poached power from the at least one first poachee server to a first poacher server thereby increasing the thermal multiplier of the first poacher server to around the first power state, wherein the first poacher server is in relatively close proximity to the at least one first poachee server. 23. The system according to claim 22, wherein the resource manager is further configured to determine servers having thermal multipliers below a second power state and above the first power state and to poach power from at least one second poachee server having a thermal multiplier below the second power state and above the first power state and to give the poached power from the at least one second poachee server to a second poacher server thereby increasing the thermal multiplier of the second poacher server to the around the second power state, wherein the second poacher server is in relatively close proximity to the at least one second poachee server. 24. The system according to claim 23, wherein the resource manager is further configured to update the thermal multipliers of the at least one second poachee server by a second poaching factor, wherein the second poaching factor is a function of a distance between the second poacher server and the at least one second poachee server, and wherein the resource manager is further configured to mark the least one second poachee server as having been poached. 25. The system according to claim 20, wherein the resource manager is further configured to calculate target power consumption levels for a plurality of racks in a row of racks and to determine the one of the plurality of power states to assign to the servers contained in the plurality of racks. 26. The system according to claim 25, wherein the resource manager is further configured to compare a cumulative power budget for the racks to an assignable threshold power level and to assign the threshold power levels to those racks having cumulative power budgets that exceed the assignable threshold power level. 27. The system according to claim 26, wherein the resource manager is further configured to calculate said thermal multipliers for the servers based upon the calculated target power consumption levels for a plurality of racks, to create power budgets for the servers contained in the plurality or racks and to assign said power states to the servers based upon a comparison of the thermal multipliers and the power states. 28. The system according to claim 25, wherein the resource manager is farther configured to give the poached power from the at least one first poachee server to a first poacher server thereby increasing the thermal multiplier of the first poacher server to around the first power state, wherein the first poacher server is in relatively close proximity to the at least one first poachee server. 29. The system according to claim 28, wherein the resource manager is further configured to determine servers having thermal multipliers below a second power state and above the first power state and to poach power from at least one second poachee server having a thermal multiplier below the second power state and above the first power state and to give the poached power from the at least one second poachee server to a second poacher server thereby increasing the thermal multiplier of the second poacher server to the around the second power state, wherein the second poacher server is in relatively close proximity to the at least one second poachee server. 30. A data center having a system far power distribution among servers, said data center comprising: means for detecting temperatures of airflow exhausted from the servers; means for detecting a temperature of die airflow supplied from a CRAC unit; means for controlling power distribution in the data center, said means for controlling comprising means for calculating thermal multipliers of the servers, means for discretizing the thermal multipliers of the servers to one of a plurality of server power states based upon geographic locations of the servers, and means for assigning one of the plurality of server power states to the servers based upon the discretization of the thermal multiplier; and means for determining servers having thermal multipliers below a first power state and to poach power from at least one first poachee server having a thermal multiplier below the first power suite. 31. A tangible computer readable storage medium on which is embedded one or more computer programs, said one or more computer programs implementing a method of distributing power among servers, said one or more computer programs comprising a set of instructions for; calculating thermal multipliers of the servers; discretizing the thermal multipliers of the servers to one of a plurality of server power states based upon geographic locations of the servers; assigning one of the plurality of server power states to the servers based upon the discretization of the thermal multipliers; determining servers having thermal multipliers close to a first power state; and for a first poacher server, poaching power from at least one first poachee server having a thermal multiplier below the first power state.
연구과제 타임라인
LOADING...
LOADING...
LOADING...
LOADING...
LOADING...
이 특허에 인용된 특허 (1)
Bradley, David John; Harper, Richard Edwin; Hunter, Steven Wade, Method for managing power consumption of multiple computer servers.
Akiyama, Kazuhito; Furuichi, Sanehiro; Tada, Masami, Changing operating state of a network device on a network based on a number of users of the network.
Weber, Wolf-Dietrich; Fan, Xiaobo; Barroso, Luiz Andre, Data center load monitoring for utilizing an access power amount based on a projected peak power usage and a monitored power usage.
Fan, Xiaobo; Hennecke, Mark D.; Heath, Taliver Brooks, Method of correlating power in a data center by fitting a function to a plurality of pairs of actual power draw values and estimated power draw values determined from monitored CPU utilization of a statistical sample of computers in the data center.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.