We consider a large-scale service system where incoming tasks have to be instantaneously dispatched to one out of many parallel server pools. The user-perceived performance degrades with the number of concurrent tasks and the dispatcher aims at maximizing the overall quality of service by balancing the load through a simple threshold policy. We demonstrate that such a policy is optimal on the fluid and diffusion scales, while only involving a small communication overhead, which is crucial for large-scale deployments. In order to set the threshold optimally, it is important, however, to learn the load of the system, which may be unknown. For that purpose, we design a control rule for tuning the threshold in an online manner. We derive conditions that guarantee that this adaptive threshold settles at the optimal value, along with estimates for the time until this happens. In addition, we provide numerical experiments that support the theoretical results and further indicate that our policy copes effectively with time-varying demand patterns. Summary of Contribution: Data centers and cloud computing platforms are the digital factories of the world, and managing resources and workloads in these systems involves operations research challenges of an unprecedented scale. Due to the massive size, complex dynamics, and wide range of time scales, the design and implementation of optimal resource-allocation strategies is prohibitively demanding from a computation and communication perspective. These resource-allocation strategies are essential for certain interactive applications, for which the available computing resources need to be distributed optimally among users in order to provide the best overall experienced performance. This is the subject of the present article, which considers the problem of distributing tasks among the various server pools of a large-scale service system, with the objective of optimizing the overall quality of service provided to users. A solution to this load-balancing problem cannot rely on maintaining complete state information at the gateway of the system, since this is computationally unfeasible, due to the magnitude and complexity of modern data centers and cloud computing platforms. Therefore, we examine a computationally light load-balancing algorithm that is yet asymptotically optimal in a regime where the size of the system approaches infinity. The analysis is based on a Markovian stochastic model, which is studied through fluid and diffusion limits in the aforementioned large-scale regime. The article analyzes the load-balancing algorithm theoretically and provides numerical experiments that support and extend the theoretical results.
more »
« less
Data for: The 30 Doradus Molecular Cloud at 0.4 pc Resolution with ALMA: Physical Properties and the Boundedness of CO-emitting Structures
12CO and 13CO emission maps of the 30 Doradus molecular cloud in the Large Magellanic Cloud, obtained with the Atacama Large Millimeter/submillimeter Array (ALMA) during Cycle 7. See the associated article in the Astrophysical Journal, and README file, for details. Please cite the article if you use these data.
more »
« less
- Award ID(s):
- 2009849
- NSF-PAR ID:
- 10443120
- Author(s) / Creator(s):
- ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; more »
- Publisher / Repository:
- University of Illinois at Urbana-Champaign
- Date Published:
- Edition / Version:
- 1
- Subject(s) / Keyword(s):
- ["Radio astronomy"]
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
more » « less
Abstract This article presents a new deep convective parameterization that determines cloud characteristics based on a specified cloud size distribution. The vertical profiles of cloud properties are determined by analytic equations, which formulate entrainment with an inverse relationship to cloud width. In line with recent studies of large eddy simulations (LES), cloud widths are assumed to be constant with height and vertical mass flux (
M ) characteristics of the clouds are therefore regulated by the vertical velocity profile. The parameterization is configured to work with existing cloud baseM closure formulations, with the closure predicting the total cloud area rather than the cloud baseM directly. Analytic formula are also used to connect the vertical wind shear magnitude to the cloud size distribution, wherein larger shear magnitudes result in more numerous large updrafts than weaker shear magnitudes, which is in line with recent research results. The parameterization is compared against 10 deep convective LES with varying thermodynamic and vertical wind shear profiles. Results show dramatic improvements in the prediction of normalizedM , detrainment, and the properties of detrained air over the existing Zhang and McFarlane (1995) scheme. In particular, the new model is able to correctly portray the transition from a bottom‐heavyM profile in weakly sheared environments, to a top‐heavyM profile in strongly sheared environments. -
In this article, we survey existing academic and commercial efforts to provide Field-Programmable Gate Array (FPGA) acceleration in datacenters and the cloud. The goal is a critical review of existing systems and a discussion of their evolution from single workstations with PCI-attached FPGAs in the early days of reconfigurable computing to the integration of FPGA farms in large-scale computing infrastructures. From the lessons learned, we discuss the future of FPGAs in datacenters and the cloud and assess the challenges likely to be encountered along the way. The article explores current architectures and discusses scalability and abstractions supported by operating systems, middleware, and virtualization. Hardware and software security becomes critical when infrastructure is shared among tenants with disparate backgrounds. We review the vulnerabilities of current systems and possible attack scenarios and discuss mitigation strategies, some of which impact FPGA architecture and technology. The viability of these architectures for popular applications is reviewed, with a particular focus on deep learning and scientific computing. This work draws from workshop discussions, panel sessions including the participation of experts in the reconfigurable computing field, and private discussions among these experts. These interactions have harmonized the terminology, taxonomy, and the important topics covered in this manuscript.more » « less
-
In this article, we survey existing academic and commercial efforts to provide Field-Programmable Gate Array (FPGA) acceleration in datacenters and the cloud. The goal is a critical review of existing systems and a discussion of their evolution from single workstations with PCI-attached FPGAs in the early days of reconfigurable computing to the integration of FPGA farms in large-scale computing infrastructures. From the lessons learned, we discuss the future of FPGAs in datacenters and the cloud and assess the challenges likely to be encountered along the way. The article explores current architectures and discusses scalability and abstractions supported by operating systems, middleware, and virtualization. Hardware and software security becomes critical when infrastructure is shared among tenants with disparate backgrounds. We review the vulnerabilities of current systems and possible attack scenarios and discuss mitigation strategies, some of which impact FPGA architecture and technology. The viability of these architectures for popular applications is reviewed, with a particular focus on deep learning and scientific computing. This work draws from workshop discussions, panel sessions including the participation of experts in the reconfigurable computing field, and private discussions among these experts. These interactions have harmonized the terminology, taxonomy, and the important topics covered in this manuscript.more » « less
-
more » « less
Abstract When the electric field below a thunderstorm or other electrified cloud is around 10 kV/m, it is sometimes possible to initiate (“trigger”) an upward‐propagating lightning‐leader by launching a rocket that uncoils a wire from the ground. The triggered leader propagates upward from the tip of the wire lifted by the rocket. When the channel is hot enough, a flash is visible. Triggering is common when the leader carries positive charge, but not when it carries negative charge. This article is about four flashes consisting of triggered negative leaders that branched into low‐altitude regions of positive cloud charge over Langmuir Laboratory in central New Mexico. Measurements of current and the locations of leader channels are available for three of the four flashes. Some current pulses at the ground for Flash 2 originated at negative leader steps more than 3 km away, which is a greater distance than has been reported from video measurements. Flashes 3 and 4 propagated only into thunderstorm lower positive charge, and the average lightning‐charge densities inside the volumes occupied by these two flashes are remarkably close. Our best estimate of density for Flashes 3 and 4 lies between −4.2 and −1.8 C/km3, which is compatible with the large spread in cloud‐charge densities derived from instruments carried on airplanes or balloons into low positive regions in thunderstorms.
