Temporal Workload Research Articles

Exploring the impact of word prediction assistive features on smartphone keyboards for blind users

Assistive technologies have been developed to enhance blind users’ typing performance, focusing on speed, accuracy, and effort reduction. One such technology is word prediction software, designed to minimize keystrokes required for text input. This study investigates the impact of word prediction on typing performance among blind users using an on-screen QWERTY keyboard. We conducted a comparative study involving eleven blind participants, evaluating both standard QWERTY input and word prediction-assisted typing. Our findings reveal that while word prediction slightly improves typing speed, it does not enhance typing accuracy and increases both physical and temporal workload compared to the default keyboard. We conclude with recommendations for improving word prediction systems, including more efficient editing methods and the integration of voice pitch variations to aid error recognition.

The Burden of Administrative Household Labor—Measuring Temporal Workload, Mental Workload, and Satisfaction

This research paper investigates the temporal and mental workload as well as work satisfaction regarding bureaucratic, administrative household labor, with a focus on socio-demographic differences. The study utilizes a paid online survey with 617 socio-demographically distributed participants. The results show significant differences in the temporal workload of different chore categories and in the quality of work, whereby satisfaction and mental workload are examined. In addition, the influences of gender, age, and education are analyzed, revealing differences in temporal and mental workload as well as work satisfaction. Our findings confirm prevailing literature showing that women have lower work satisfaction and a higher workload. In addition, we also discovered that younger people and groups of people with higher incomes have a higher level of satisfaction and a higher workload. In our study, a perceived high mental workload does not necessarily go hand in hand with a low level of satisfaction. This study contributes to the understanding of the bureaucratic burden on adults in their households and the variety of activities to manage private life.

Multi-assistance systems in manufacturing - a user study evaluating multi-criteria impact in a high-mix low-volume assembly setting

Understanding how skilled workers interact with assistance systems in manufacturing and how they experience the factory environment is fundamental to modeling human interaction and optimizing the processes correctly. This paper investigates humans’ behaviors and perceived experiences while interacting with cognitive and physical assistance systems. To enable decisions about the combined use of more than one assistance system within a manufacturing process, comprehensive and comparable knowledge about the impact of applications on productivity and human factors is needed. A multidimensional evaluation model with a mixed-methods approach was developed and applied in a user study. In 300 run-throughs in six different scenarios with skilled workers and students, a questionnaire on human factors was completed after finishing the task. Furthermore, productivity and quality were measured during the study. A comparison between skilled workers and students demonstrated that the usability score of all assistance systems was rated higher among the students. The students rated the ergonomics aspects better for five out of six scenarios. Results show higher compatibility with values and experiences in all investigated combinations for skilled workers than for students. Considering the collected data among the skilled workers, the overall compatibility with experience and values of multi-assistance system scenarios was more positive than in the single-assistance system scenarios. Our results show no significant differences in ergonomics, mental, physical, and temporal workload between single and multi-assistance system settings. With 150 run-throughs of industrial professionals and campus recruits each, the survey joins only two studies with more than 100 participants. To the authors' knowledge, it is the first systematic multi-criteria evaluation for the combined use of several (cognitive and physical) industrial assistance systems. The results help to ease practitioners’ evaluation of technical support systems in manufacturing with an emphasis on multi-criteria evaluation and the consideration of interconnected (cognitive and physical) assistance systems. Furthermore, the results contribute to further research in human–machine interaction and its impact on productivity and human factors as they show potentials and prospective challenges of the implementation and application of multiple assistance systems.

Generating balanced workload allocations in hospitals

As pressure on healthcare systems continues to increase, it is becoming more and more important for hospitals to properly manage the high workload levels of their staff. Ensuring a balanced workload allocation between various groups of employees in a hospital has been shown to contribute considerably towards creating sustainable working conditions. However, allocating work to different organizational units in a fair manner is not straightforward when it involves complex decision-making processes. In this paper we set out to balance the workload of heterogeneous hospital wards by optimizing the patient admission scheduling problem. Given the multi-period nature of patient admission scheduling, we introduce a new equity objective that captures both spatial (between hospital wards) and temporal (between days in the planning period) workload balancing. The resulting bi-objective problem is solved using an exact criterion space search algorithm. Our computational study employs problem instances that have been generated based on real-world data. The results demonstrate how spatially and temporally balanced workload allocations can be generated by minimizing the proposed equity objective. Moreover, we analyze sets of nondominated solutions to gain various insights into the trade-off between schedule cost and workload balance.

SPCIM: Sparsity-Balanced Practical CIM Accelerator With Optimized Spatial-Temporal Multi-Macro Utilization

Compute-in-memory (CIM) is a promising technique that reduces data movement in neural network (NN) acceleration. To achieve higher efficiency, some recent CIM accelerators exploit NN sparsity based on CIM’s small-grained operation unit (OU) feature. However, new problems arise in a practical multi-macro accelerator: The mismatch between workload parallelism and CIM macro organization causes spatial under-utilization; The multiple macros’ different computation time leads to temporal under-utilization. To solve the under-utilization problems, we propose a Sparsity-balanced Practical CIM accelerator (SPCIM), including optimized dataflow and hardware architecture design. For the CIM dataflow design, we first propose a reconfigurable cluster topology for CIM macro organization. Then we regularize weight sparsity in the OU-height pattern and reorder the weight matrix based on the sparsity ratio. The cluster topology can be reshaped to match workload parallelism for higher spatial utilization. Each CIM cluster’s workload is dynamically rebalanced for higher temporal utilization. Our hardware architecture supports the proposed dataflow with a spatial input dispatcher and a temporal workload allocator. Experimental results show that, compared with the baseline sparse CIM accelerator that suffers from spatial and temporal under-utilization, SPCIM achieves <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2.94\times $ </tex-math></inline-formula> speedup and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$2.86\times $ </tex-math></inline-formula> energy saving. The proposed sparsity-balanced dataflow and architecture are generic and scalable, which can be applied to other CIM accelerators. We strengthen two state-of-the-art CIM accelerators with the SPCIM techniques, improving their energy efficiency by <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$1.92\times $ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$5.59\times $ </tex-math></inline-formula> , respectively.

Benefits and Risks of Teleworking from Home: The Teleworkers’ Point of View

Using a qualitative research-based approach, this study aimed to understand (i) the way home-based teleworkers in France perceive and organize their professional activities and workspaces, (ii) their teleworking conditions, (iii) the way they characterize the modalities and the nature of their interactions with their professional circle, and more broadly (iv) their quality of life ‘at work’. We performed a lexical and morphosyntactic analysis of interviews conducted with 28 teleworkers (working part-time or full-time from home) before the COVID-19 crisis and the associated establishment of emergency telework. Our results confirm and complement findings in the literature. Participant discourses underlined the beneficial effects of teleworking in terms of professional autonomy, flexibility, concentration, efficiency, performance, productivity, and being able to balance their professional and private lives. Nevertheless, they also highlighted the deleterious effects of teleworking on temporal workload, setting boundaries for work, work-based relationships and socio-professional integration. Despite the study limitations, our findings highlight the need for specific research-based and practical strategies to support the implementation of a sustainable telework organization in the post-COVID-19 pandemic era.

Design and Formative Evaluation of a Virtual Voice-Based Coach for Problem-solving Treatment: Observational Study.

BackgroundArtificial intelligence has provided new opportunities for human interactions with technology for the practice of medicine. Among the recent artificial intelligence innovations, personal voice assistants have been broadly adopted. This highlights their potential for health care–related applications such as behavioral counseling to promote healthy lifestyle habits and emotional well-being. However, the use of voice-based applications for behavioral therapy has not been previously evaluated.ObjectiveThis study aimed to conduct a formative user evaluation of Lumen, a virtual voice-based coach developed as an Alexa skill that delivers evidence-based, problem-solving treatment for patients with mild to moderate depression and/or anxiety.MethodsA total of 26 participants completed 2 therapy sessions—an introductory (session 1) and a problem-solving (session 2)—with Lumen. Following each session with Lumen, participants completed user experience, task-related workload, and work alliance surveys. They also participated in semistructured interviews addressing the benefits, challenges and barriers to Lumen use, and design recommendations. We evaluated the differences in user experience, task load, and work alliance between sessions using 2-tailed paired t tests. Interview transcripts were coded using an inductive thematic analysis to characterize the participants’ perspectives regarding Lumen use.ResultsParticipants found Lumen to provide high pragmatic usability and favorable user experience, with marginal task load during interactions for both Lumen sessions. However, participants experienced a higher temporal workload during the problem-solving session, suggesting a feeling of being rushed during their communicative interactions. On the basis of the qualitative analysis, the following themes were identified: Lumen’s on-demand accessibility and the delivery of a complex problem-solving treatment task with a simplistic structure for achieving therapy goals; themes related to Lumen improvements included streamlining and improved personalization of conversations, slower pacing of conversations, and providing additional context during therapy sessions.ConclusionsOn the basis of an in-depth formative evaluation, we found that Lumen supported the ability to conduct cognitively plausible interactions for the delivery of behavioral therapy. Several design suggestions identified from the study including reducing temporal and cognitive load during conversational interactions, developing more natural conversations, and expanding privacy and security features were incorporated in the revised version of Lumen. Although further research is needed, the promising findings from this study highlight the potential for using Lumen to deliver personalized and accessible mental health care, filling a gap in traditional mental health services.

Cloud data center participation in smart demand response programs for energy cost minimisation

Demand Response Programs (DRPs) in Smart Grid (SG) are designed to encourage consumers to shift their loads to regions and hours with less load stress by alternating the price of electricity. On the other hand, with the emergence of cloud computing, the demand for cloud data centres increases resulting in high energy costs. In this paper, we address the problem of cloud data centre demand response in an SG area in presence of utility companies which compete to attract more customers. We formulate two optimization problems for cloud and data centres to minimise their costs by changing their energy demand. These problems are shown to be convex and can be readily solved by standard convex optimization techniques. We also propose an algorithm for cloud and data centres to participate in DRPs by concurrently performing regional and temporal workload management. The uncertainty of renewable energy generation in data centres is treated by training a multilayer perceptron to predict the generated energy. Numerical results show that the proposed algorithm outperforms other existing algorithms in terms of the energy cost. In addition, our algorithm flattens the energy demand profile of utility companies and balances the electric load across different locations.

183 Preliminary Results of Assessing Cognitive Load During Procedure Training

During simulated teaching of a medical procedural skill, learners must develop, refine, and activate complex mental models for cognitive and psychomotor tasks. These processes may burden the learner and contribute to cognitive load. Augmented reality (AR) systems can expand access to training by facilitating remote interactions. However, the technology itself can impact cognitive workload and affect learning. In advance of deploying an AR training system, we seek to define the workload associated with the simulation environment. This pilot study uses NASA-TLX, a subjective workload rating tool, and SIM-TLX, a derivative designed to evaluate workload in simulated environments, to measure contributions to workload in instructor/learner dyads during in person, Ultrasound Guided Central Venous Catheter training. We postulate that the SIM TLX in the in-person training setting will result in low environment-based workload ratings. A prospective observational study was conducted after IRB approval. Five learners with minimal US-CVC experience were paired with 5 physician instructors. Each trainee watched an instructional video prior to hands on, in-person skills practice using a commercially available US CVC simulator. Both the instructor and trainee completed post session NASA TLX and SIM TLX surveys. Workload ratings were compared using paired, 2 tailed t tests to identify individual workload item contributions. Average results of the overall and workload subscales are presented in Figure 1. Learners reported a non-significantly higher overall workload than instructors. In the NASA TLX, learners rated Mental Workload higher than Temporal Workload, Frustration and Effort; and Performance and Effort higher than Frustration as components of the overall workload (Paired 2 tailed t-test, p ≤0.05), In the SIM TLX, learners rated Mental Workload higher than Temporal Workload, Frustration, Situational Stress, Distraction, and Perceptual Strain; Task Completion higher than Temporal Stress, Situational Stress, Distraction and Perceptual Strain; and Task Control higher than Situational Stress or Temporal Workload. Instructors rated Mental Workload higher than Physical Workload in both surveys, and higher than Distraction and Perceptual Strain in the SIM TLX. In the in-person training environment, both instructors and trainees identified Mental Workload and Task Complexity as greater contributors to workload than Perceptual Strain and Distraction subscales in the SIM TLX, a finding not reflected in the NASA TLX subscales. We conclude that the simulation environment during in-person training contributed minimally to cognitive load. Further workload evaluation in AR and other training environments will help to delineate the magnitude of environmental contributions to workload during training.

RELATIONSHIP BETWEEN SURGEON'S PHYSIOLOGICAL AND KINEMATIC PARAMETERS AND THE QUALITY OF SURGICAL PERFORMANCE AND WORKLOAD IN LAPAROSCOPIC TRAINING

Abstract INTRODUCTION This study presents the first steps and results towards the development of a system for predicting the quality of surgical performance and workload in laparoscopic training. MATERIAL AND METHODS Surgeons wore a smartwatch which recorded their heart rate and hand motion during each laparoscopic procedure. Data was then correlated with both the Surgery Task Load Index (SURG-TLX) subjective questionnaire and the Global Operational Assessment of Laparoscopic Skills (GOALS) objective evaluation tool. Statistical analysis was conducted in order to study the relationship between parameters and to compare the results according to the surgeons’ level of experience. RESULTS Nine laparoscopic surgeons participated in this study, five gynecologists and four digestive surgeons. Gynecological surgeons showed a positive correlation between their level of experience, heart rate, hand motion, and GOALS score, except for bimanual dexterity. The reduction in the variability of hand accelerations led to improved tissue handling. Digestive surgeons showed a negative correlation between their level of experience and the reported temporal demand and complexity of the procedure. Novice digestive surgeons reported increased workload during surgery, mainly in mental, physical, and temporal workload. In both surgical specialties, it was observed that reduction in surgical workload was associated with improved GOALS score. CONCLUSIONS Preliminary results showed that bimanual dexterity is independent of the experience of gynecological surgeons. During laparoscopic surgical procedures, reduced variability in hand accelerations leads to better tissue manipulation. In addition, in gynecological and upper gastrointestinal procedures, reduced surgical workload is associated with improved surgical performance.

Dynamic Scheduling for Stochastic Edge-Cloud Computing Environments Using A3C Learning and Residual Recurrent Neural Networks

The ubiquitous adoption of Internet-of-Things (IoT) based applications has resulted in the emergence of the Fog computing paradigm, which allows seamlessly harnessing both mobile-edge and cloud resources. Efficient scheduling of application tasks in such environments is challenging due to constrained resource capabilities, mobility factors in IoT, resource heterogeneity, network hierarchy, and stochastic behaviors. Existing heuristics and Reinforcement Learning based approaches lack generalizability and quick adaptability, thus failing to tackle this problem optimally. They are also unable to utilize the temporal workload patterns and are suitable only for centralized setups. However, asynchronous-advantage-actor-critic (A3C) learning is known to quickly adapt to dynamic scenarios with less data and residual recurrent neural network (R2N2) to quickly update model parameters. Thus, we propose an A3C based real-time scheduler for stochastic Edge-Cloud environments allowing decentralized learning, concurrently across multiple agents. We use the R2N2 architecture to capture a large number of host and task parameters together with temporal patterns to provide efficient scheduling decisions. The proposed model is adaptive and able to tune different hyper-parameters based on the application requirements. We explicate our choice of hyper-parameters through sensitivity analysis. The experiments conducted on real-world data set show a significant improvement in terms of energy consumption, response time, Service-Level-Agreement and running cost by 14.4, 7.74, 31.9, and 4.64 percent, respectively when compared to the state-of-the-art algorithms.

Spinning gold from straw - evaluating the flexibility of data centres on power markets

Data centres have been the focus of research as candidates for demand response or other demand side management programs for quite some time. However, a complete framework optimising data centre demand response is still missing. This is due to the complexity of integrating more than one power flexibility technology and more than one market for power flexibility. In the presented work, this challenge is solved by creating a microeconomics inspired optimisation approach that takes the view of a data centre offering power flexibility as a ’product’ to explicit and/or implicit demand response power flexibility markets. This generic framework is then instantiated in a linear optimisation problem that optimises the power flexibility of a German High Performance Computing Centre on a set of different power flexibility markets in Germany. It is consequently shown that, under the described scenario, frequency scaling should be preferred to temporal workload shifting and that the EPEX day ahead market is the most beneficial power flexibility market.

Simulation-based training of critical events during cardiopulmonary bypass: importance of a critical events checklist.

Evaluation of critical events training for clinical perfusionists is necessary to improve this educational approach. Critical events checklists are effective in reducing clinical complications, but should be tested in a simulation environment first. Individual behavior and stress response of clinical perfusionists during simulated critical events on cardiopulmonary bypass have not been evaluated yet. This study focuses on the evaluation of critical events training and critical events checklists in simulated cardiopulmonary bypass. A total of 19 clinical perfusionists from a single hospital took part in two simulated critical event scenarios. Clinical perfusionist behavior and physiological responses were recorded using eye tracking, heart rate variability, video, and audio. In addition, workloads were determined and participants were interviewed. Relevant areas of interest were identified for each simulation phase. During critical event detection and subsequent decision-making, areas of interest hits and fixation durations varied with the use of a critical events checklist. Times to decision were shorter, decision quality was higher, and temporal workload was increased when the checklist was used. Evaluation of selected heart rate variability measures revealed a good correlation with pupil diameters. Evaluation of critical events during simulated cardiopulmonary bypass shows that the scenario is realistic and relevant for clinical practice. Integrating a critical events checklist improves the probability of correct decision-making and shortens the correct decision time. Temporal workload is increased when using a checklist. Eye tracking and heart rate variability are well suited to evaluate participants' behaviors and stress levels. All participants welcomed simulation training for critical incidents.

System Dialog Message and Its Effect on Task Switching by Age

ABSTRACT System dialog messages are often troublesome for mobile users. In this study, behavioral experiments were conducted on 18 younger (age; 24.7 ± 1.7) and 18 older (63.8 ± 2.0) age groups using mobile devices. In the experiment, system dialog messages were popped up on a smartphone as an interruption in the task switching process, and which attributes of the dialog messages (the amount of jargon and length of the message) significantly affected the performance of the whole task switching process was observed. The results showed that the younger participants did not show significant differences in their task switching performance according to the attributes of the dialog message. However, the older participants showed increased task performance time for the interruption task with extensive jargon. Furthermore, the older participants also revealed difficulty transferring information as they resumed the primary task just after processing the interrupting dialog messages, suggested by correlations between temporal and subjective workload responses. Based on the analysis results, we proposed some guidelines to design system dialog messages for older mobile device users.

The Influence of Monochronic/Polychronic Time Orientation on Temporal Demand and Subjective Mental Workload

The sensitivity of mental workload measures is influenced by cultural and individual factors. One individual factor that is hypothesized to influence mental workload is time orientation. The aim of this study is to observe the influence of time orientation on temporal demand and subjective mental workload. One hundred and two participants representing three different time orientations, namely monochronic, neutral, and polychronic orientations, assessed using the Modified Polychronic Attitude Index 3 (MPAI3), voluntarily participated in this study. Participants were instructed to complete a search and count task in four different conditions with varying degrees of difficulty. Mental workload was assessed using subjective (NASA-TLX) and objective (heart rate variability, or HRV) methods and analyzed for each condition. The results show that, with comparable performance and comparable HRV, monochronic participants show higher sensitivity than neutral or polychronic participants in subjective mental workload, particularly the temporal demand dimension. The implications are discussed.

Evaluation of attention, perception, and stress levels of clinical cardiovascular perfusionists during cardiac operations: a pilot study.

Performing cardiopulmonary bypass is a complex task which involves evaluating visual input from patient monitors and technical parameters displayed at the heart-lung machine console as well as reacting to other sensory input. Only few studies are available concerning the competency requirements for clinical cardiovascular perfusionists, including attention, perception, and coping with mental stress. This study aims at evaluating attention, perception, and stress levels of clinical cardiovascular perfusionists during cardiopulmonary bypass. Nine clinical cardiovascular perfusionists voluntarily offered to participate in the study. Participants were asked to wear Tobii 2 eye-tracking glasses throughout the procedures. Specific time points were analyzed (cardiopulmonary bypass on, initial cardioplegia delivery, steady state, cross-clamp off, and weaning from cardiopulmonary bypass). Data acquisition was supplemented by participants' self-evaluation regarding their stress levels and by National Aeronautics and Space Administration Task Load Index (NASA TLX) questionnaires. Seven datasets were sufficient to be evaluated. The clinical cardiovascular perfusionists' professional experience ranged from 0.5 to 24 years. Evaluation of eye-tracking data revealed large variations in areas of interest hits, fixation, and dwell times. Across all phases, the venous reservoir, mean arterial pressure, arterial pump display, cardioplegia control, and data management system received the highest levels of attention. Pupil diameter measurements increased at start of cardiopulmonary bypass, cardioplegia delivery, and weaning off, but returned to base level during steady state. Clinical cardiovascular perfusionists' self-evaluation showed that subjective stress level was highest at the start and the end of the procedure. NASA TLX questionnaires revealed medium-to-high mental and temporal workloads, but low physical workloads. Performance, effort, and frustration indices showed medium workloads. During cardiopulmonary bypass, perfusionists are subjected to stress. Peak stress levels were highest during start and end of cardiopulmonary bypass. Furthermore, visual attention and perception varied between the operative phases. Further studies are indicated to evaluate the design of heart-lung machines and stress-coping strategies during cardiopulmonary bypass.

Analysis of the Effects of Group Size and Learning on Manual Assembly Performance

On the basis of a prior experimental study, this paper performs a further analysis of the effects of group size (one to four workers) and learning (up to four repetitions per group) on manual assembly performance. More specifically, this paper aims to investigate the factors and the extent to which they affect reduced assembly time as a function of repetitions and reduced productivity per worker as a function of increasing group size. The ultimate aim of this study is to increase the understanding of how working in groups of different sizes develops through repetitions when workers are free to organize their work themselves. The results from the video-based analysis show that with a new, relatively complex product, instructions play a crucial role in learning and the losses caused by the inexperience of workers decrease rapidly through repetitions. Unequal temporal workloads between workers in larger groups increase idleness and cause a significant loss of productivity. The findings presented in this paper give insights for industrial managers when assigning workers to products in variable assembly production of highly customized products.

Time Pressure and Teams: Contextual Factors Influencing Task Experience when Using a Cognitive Aid

Guidelines are needed to help designers create cutting-edge cognitive aids for complex procedures, particularly when those aids support non-expert operators. To create guidelines, the factors that might interact with the aid design need be explored with non-experts, such as time pressure and the number of operators. We manipulated time pressure and number of team members in setting up a medical ventilator while using a cognitive aid. We measured dependence on the aid, performance, subjective workload, and team dynamics to better understand their influence on the use of a cognitive aid to accomplish a complex task by non-expert operators. Individuals reported a significantly higher mental workload level than teams and those under time pressure reported higher temporal workload. These data can contribute to design guidelines.

Temporal workload in economic organizations: A hidden non-linear condition of economic efficiency

A temporal workload model is introduced to identify the relationship between the work time and economic performance of the activities conducted by a human agent in the context of an economic organization. The model's novelty derives from the account of time perception and its consequent cognitive time distortion, the latter being understood as a discrepancy between physical and cognitive time. Current praxis, both theoretical and empirical, assumes only physical time. This assumption is challenged here through the inclusion of time perception and cognitive time distortion in estimating the temporal workload of an economic agent. This inclusion enables a novel comprehension of frequent operational challenges, such as work delays, human stress, output quality issues, and economic inefficiencies. The main contribution to the literature is a specification of a new condition that governs the performance of any economic organization where human agents conduct time assessments.

Performance and Cost Considerations for Providing Geo-Elasticity in Database Clouds

Online applications that serve global workload have become a norm and those applications are experiencing not only temporal but also spatial workload variations. In addition, more applications are hosting their backend tiers separately for benefits such as ease of management. To provision for such applications, traditional elasticity approaches that only consider temporal workload dynamics and assume well-provisioned backends are insufficient. Instead, in this article, we propose a new type of provisioning mechanisms—geo-elasticity, by utilizing distributed clouds with different locations. Centered on this idea, we build a system called DBScale that tracks geographic variations in the workload to dynamically provision database replicas at different cloud locations across the globe. Our geo-elastic provisioning approach comprises a regression-based model that infers database query workload from spatially distributed front-end workload, a two-node open queueing network model that estimates the capacity of databases serving both CPU and I/O-intensive query workloads and greedy algorithms for selecting best cloud locations based on latency and cost. We implement a prototype of our DBScale system on Amazon EC2’s distributed cloud. Our experiments with our prototype show up to a 66% improvement in response time when compared to local elasticity approaches.