An official website of the United States government
Due to a growing interest in deep learning applications [5], compute-intensive and long-running (hours to days) training jobs have become a significant component of datacenter workloads. A large fraction of these jobs is often exploratory, with the goal of determining the best model structure (e.g., the number of layers and channels in a convolutional neural network), hyperparameters (e.g., the learning rate), and data augmentation strategies for the target application. Notably, training jobs are often terminated early if their learning metrics (e.g., training and validation accuracy) are not converging, with only a few completing successfully. For this motivating application, we consider the problem of scheduling a set of jobs that can be terminated at predetermined checkpoints with known probabilities estimated from historical data. We prove that, in order to minimize the time to complete the first K successful jobs on a single server, optimal scheduling does not require preemption (even when preemption overhead is negligible) and provide an optimal policy; advantages of this policy are quantified through simulation. Related Work. While job scheduling has been investigated extensively in many scenarios (see [6] and [2] for a survey of recent result), most policies require that the cost of waiting times of each job be known at scheduling time; in contrast, in our setting the scheduler does not know which job will be the K-th successful job, and sojourn times of subsequent jobs do not contribute to the target metric. For example, [4, 3] minimize makespan (i.e., the time to complete all jobs) for known execution times and waiting time costs; similarly, Gittins index [1] and SR rank [7] minimize expected sojourn time of all jobs, i.e., both successfully completed jobs and jobs terminated early. Unfortunately, scheduling policies not distinguishing between these two types of jobs may favor jobs where the next stage is short and leads to early termination with high probability, which is an undesirable outcome in our applications of interest.
more »
« less
Good jobs, scam jobs: Detecting, normalizing, and internalizing online job scams during the COVID-19 pandemic
Good jobs that allow remote work have enabled white-collar professionals to stay home during COVID-19, but for precarious workers, online advertisements for work-from-home employment are often scams. In this article, based on in-depth interviews conducted between April and July 2020 with nearly 200 precarious workers, we find that precarious workers regularly encountered fraudulent job advertisements via digital media. Drawing on Swidler’s concepts of the cultural tool kit and cultural logic, we find that in this time of uncertainty, workers defaulted to the focus on personal responsibility that is inherent in insecurity culture. Following the cultural logic of personal responsibility, job seekers did not place blame on job search websites for allowing the scams to be posted, but normalized the situation, deploying a scam detection repertoire in response. In addition, the discovery that advertised “good jobs” are often scams affecting workers’ desire to continue job hunting and perceptions of potential future success.
more »
« less
- Award ID(s):
- 2029924
- PAR ID:
- 10548118
- Publisher / Repository:
- SAGE Publications
- Date Published:
- Journal Name:
- New Media & Society
- Volume:
- 24
- Issue:
- 7
- ISSN:
- 1461-4448
- Format(s):
- Medium: X Size: p. 1591-1610
- Size(s):
- p. 1591-1610
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
The future of work is ambiguous at best. Despite widespread shifts to remote/hybrid work during the COVID-19 lockdown, there is a paucity of knowledge about changing job conditions in tandem with different work locales. Is the move to remote/hybrid work a disrupter or accentuator of existing norms and inequalities? Drawing on nationally representative, four-wave panel survey data (October 2020 to April 2022) collected from U.S. workers who spent at least some time working from home since the pandemic onset, we examine effects of within-person changes in where respondents work on changes in job conditions (psychological job demands, job control, coworker support, and monitoring). Estimates from fixed-effects models show that, compared with returning to working at work, ongoing remote and moving to hybrid work lead to greater reductions in psychological job demands, especially among older women and men. Black and Hispanic women moving back to the office experience the greatest loss of decision latitude and schedule control. While white workers see increased coworker support when returning to the office, returning Black and Hispanic men report a decline in coworker support. Family caregivers’ job conditions do not improve whether remote/hybrid or returning to work. Qualitative data collected from Amazon Mechanic Turk illuminate mechanisms leading to salutary effects of remote work, but also the stress of combining jobs with family carework.more » « less
-
The Covid-19 pandemic has greatly impacted the labor market and given rise to the Great Resignation. Drawing on a mixed methods panel study of 199 precarious and gig-based workers, we analyze how a changing conception of free time during the Covid-19 pandemic led low-wage service workers to seek more fulfilling careers. Whereas most workers initially perceived free time in terms of opportunity costs, they later reconceived this time as enabling an investment in personal growth, moving from “spending time” making money to “investing time” in themselves. This shift in temporal experience is expressed through the adoption of a “work passion” logic and “pandemic epiphanies” that motivated respondents to seek self-affirming and potentially more lucrative work opportunities.more » « less
-
Cloud computing today is dominated by multi-server jobs. These are jobs that request multiple servers simultaneously and hold onto all of these servers for the duration of the job. Multi-server jobs add a lot of complexity to the traditional one-server-per-job model: an arrival might not fit'' into the available servers and might have to queue, blocking later arrivals and leaving servers idle. From a queueing perspective, almost nothing is understood about multi-server job queueing systems; even understanding the exact stability region is a very hard problem. In this paper, we investigate a multi-server job queueing model under scaling regimes where the number of servers in the system grows. Specifically, we consider a system with multiple classes of jobs, where jobs from different classes can request different numbers of servers and have different service time distributions, and jobs are served in first-come-first-served order. The multi-server job model opens up new scaling regimes where both the number of servers that a job needs and the system load scale with the total number of servers. Within these scaling regimes, we derive the first results on stability, queueing probability, and the transient analysis of the number of jobs in the system for each class. In particular we derive sufficient conditions for zero queueing. Our analysis introduces a novel way of extracting information from the Lyapunov drift, which can be applicable to a broader scope of problems in queueing systems.more » « less
-
Abstract Immigrant day laborers routinely experience exploitative behaviors as part of their employment. These day laborers perceive the exploitation they experience in the context of their immigration histories and in the context of their long-term goals for better working and living conditions. Using mixed methods, over three data collection periods in 2016, 2019 and 2020, we analyze the work experiences of immigrant day laborers in Houston and Austin, Texas. We report how workers evaluate precarious jobs and respond to labor exploitation in an informal labor market. We also discuss data from a worker rights training intervention conducted through a city-sponsored worker center. We discuss the potential for worker centers to be a convening and remediation space for workers and employers. Worker centers offer a potential space for informal intervention into wage theft and work safety violations by regulating the hiring context where day laborers meet employers.more » « less
