A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is
Heterogeneity in modern datacenters is on the rise, in hardware resource characteristics, in workload characteristics, and in dynamic characteristics (e.g., a memoryresident copy of input data). ... constraints, (c) combinatorial constraints, (d) orderless global scheduling, and (e) in situ preemption. ... Fig. 7.14 evaluates TetriSched and TetriSched-NG (with greedy scheduling instead of global), as a function of the plan-ahead window. ...doi:10.1184/r1/6721412.v1 fatcat:7tuy7kbfkffyja4ihziaciuwzm
In our experiments, POP achieves allocations within 1.5% of the optimal with orders-of-magnitude improvements in runtime compared to existing systems for cluster scheduling, traffic engineering, and load ... into a global allocation for all clients. ... Any opinions and conclusions expressed in this material are those of the authors and do not reflect the views of the NSF. ...arXiv:2110.11927v1 fatcat:3softued4vczhbg5vfe4gg5e3m
Tetrisched uses a reservation system and plans ahead by providing a job the possibility of waiting to get a busy preferred resource or fall back to less preferred options. ... Second, we plan to make Kairos handle heterogeneous CPU allocations, where different containers can be allocated different number of cores. ... It offers short tasks the possibility of completing with limited or no waiting time, even in high-utilization scenarios. ...doi:10.5075/epfl-thesis-8892 fatcat:uywgitwzfjhifieyhj2hcr27qm
n = 1) schedule with stragglers. . . . . . 20 2.9 Exemplary depiction of a BSP schedule on a heterogeneous cluster. . . 21 2.10 Exemplary depiction of micro-task schedules on a heterogeneous cluster. ... Hardware heterogeneity This section describes two major sources of hardware heterogeneity in large-scale clusters that back the cloud. ... Both variants use the same local solver algorithm (SGD), hence lSGD with H=1 is mSGD. The number of training samples processed per iteration is K × H × L in both cases. ...doi:10.5445/ir/1000117451 fatcat:ui4lxnlonjculcs7qimbo2dena