SLA Violations Research Articles

Enhanced Ant Colony Based VM Selection and Consolidation for Energy Conservation

Cloud Computing (CC) involves extensive data centers with numerous computing nodes that consume significant electrical energy. Researchers have identified high service-level agreement (SLA) violations and excessive energy consumption (EC) as major challenges in CC. Traditional approaches often focus on reducing EC but tend to overlook SLA violations, particularly when selecting Virtual Machines (VMs) from overloaded hosts. To address these issues, this paper introduces the Enhanced Ant Colony Optimization (EACO) algorithm, aims to reduce high EC and SLA violations by utilizing a unique approach where the best-performing ant explores movement patterns and identifies distances between movements. The algorithm comprises three key steps: tracking pheromone trails, updating pheromones and selecting the cities (VMs). The effectiveness of EACO was validated through simulations using CloudSim. Compared to existing techniques, EACO demonstrated a significant reduction in EC, achieving approximately 41-44% lower energy consumption than the traditional Ant Colony Optimization (ACO) algorithm when applied to Planet Lab data. This suggests that EACO offers a more efficient and stable solution for managing EC and SLA violations in cloud environments.

Fault tolerant & priority basis task offloading and scheduling model for IoT logistics

The Internet of Things (IoT) has connected millions of devices to the internet for communication and computation purposes. Due to constraints such as limited battery power, processing capabilities, and storage capacity, IoT devices often rely on remote fog nodes for task execution, a practice known as task offloading. The NP-hard nature of offloading and scheduling IoT tasks onto fog nodes poses significant challenges. Current IoT task offloading often relies on single parameters like device MIPS, RAM, or battery power. Similarly, task allocation on fog nodes frequently prioritizes VM MIPS. To address energy consumption and task failures in IoT systems, robust fault tolerance, efficient task offloading, and effective scheduling are crucial. This paper introduces a novel Fault-Tolerant, Priority-Based Task Offloading and Scheduling Model (FP-TOSM) for IoT logistics. Utilizing the Analytic Hierarchy Process (AHP) for multi-criteria decision-making, IoT tasks are prioritized. Tasks below a specified threshold are executed locally, while those within a defined range are offloaded to a fog node. Tasks exceeding this range are delegated to the cloud. The Euclidean formula is used to determine the proximity between the IoT logistics vehicle and fog nodes for offloading decisions. Dynamic re-clustering fault tolerance mechanisms are utilized to manage task failures after offloading. Successful offloaded tasks are allocated to suitable VMs in the cloud and fog nodes. The proposed strategy selects the most suitable VM for offloaded tasks using a multi-criteria decision-making process to reduce energy consumption, SLA violations, and execution costs. The model's performance is evaluated through simulations in iFogSim2. Results demonstrate reductions in energy consumption by up to 5.8 % and 10.4 %, decreases in SLA violations by up to 25.2 %, and a 16.28 % improvement in response time compared to an ACO-based model with a response time of 17.8 %. Additionally, the task failure ratio shows a significant reduction of up to 22.1 %. These findings highlight the effectiveness of FP-TOSM in enhancing efficiency and reliability within fog computing environments.

Efficient virtual machine placement in cloud computing environment using BSO-ANN based hybrid technique

Cloud computing has revolutionized the way businesses and individuals access and utilize computing resources. Efficient virtual machine placement is a critical aspect of optimizing resource utilization, reducing operational costs, energy consumption, service level agreement and minimum virtual machine migrations, execution time, and ensuring the overall performance of cloud services. This manuscript introduces a novel approach that combines the creative problem-solving capabilities of brainstorming with the computational power of Artificial Neural Networks (ANN) to address the virtual machine placement problem in cloud environments. In this study, we propose a hybrid technique that leverages the collective intelligence of human brainstorming to generate a diverse set of placement strategies. These strategies are then evaluated, refined, and optimized using an ANN model trained on historical cloud resource allocation workload logs. By integrating the human creative process with the data-driven predictive capabilities of ANN, our approach aims to overcome the limitations of traditional virtual machine placement algorithms, which often struggle to adapt to dynamic workloads and changing resource requirements. The manuscript provides a detailed description of the hybrid technique, including the process of brainstorming for generating placement strategies, data collection and preprocessing, ANN model development, and the integration of these components into an efficient placement system. We present experimental results demonstrating the effectiveness of our approach in optimizing resource allocation, improving service performance, and optimizing resource utilization, reducing energy consumption, service level agreement and minimum virtual machine migrations, reducing execution time compared to existing static, and meta-heuristic methods. The proposed Brain Storming with ANN based Hybrid Technique offers a promising solution for enhancing the efficiency of virtual machine placement in cloud computing environments. BSO-ANN outperforms the existing techniques using performance metrics (energy consumption(Kwh), execution time(ms), SLA violations, and number of migrations). It combines human ingenuity with data-driven insights to adapt to the ever-changing dynamics of cloud workloads. This manuscript contributes to the ongoing research in cloud resource management, offering a practical approach for cloud service providers and organizations to better utilize their resources and enhance the overall quality of cloud-based services. © 2012 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Global Science and Technology Forum Pte Ltd.

Efficient energy consumption in hybrid cloud environment using adaptive backtracking virtual machine consolidation

Virtualization is the process of sharing physical machine resources into multiple Virtual Machines (VMs) for the effective utilization of resources. The major issues are in handling physical resources such as high demand of resources, infrastructure issues and operational issues. Also violating Service Level Agreement (SLA) is another problem to lead the violations. In this case, VM consolidation method is used to migrate the VMs and optimize the efficient way to use the resources. So we get minimized energy consumptions. Existing methods has increased energy consumption and overhead issues. In this paper we propose adaptive backtracking methods to provide VM consolidation based on less energy consumptions. We use Adaptive Hill Climbing and Pursuit algorithms to consolidate the VMs and use the VMs with efficient energy consumption. We implement this simulations using Matlab in Hybrid Cloud Environment. The performance factors are obtained with less energy consumption and compare the results with existing methods. Overall our proposed system is achieving 95% of accuracy index with lower energy consumption such as 28%, 30% and 32% with multiple VMs consolidations.

EBWO‐GE: An innovative approach to dynamic VM consolidation for cloud data centers

SummaryCloud data centers (CDCs) have revolutionized global computing by offering extensive storage and processing capabilities. Nevertheless, the environmental impact of these processes, including their substantial energy consumption and carbon emissions, calls for implementing more efficient techniques. Efficient virtual machine (VM) consolidation is crucial in optimizing resource utilization and reducing energy consumption. Current methods for enhancing energy efficiency often lead to issues such as service level agreements (SLAs) violations and quality of services (QoS) degradation. This study presents a novel approach to host selection using a grey‐extreme (GE) machine learning model, which accurately predicts over and underutilized hosts. In addition, a VM placement technique called enhanced black widow optimization (EBWO) utilizes black widow optimization heuristic techniques and a differential evolutionary approach to optimize VM placement. The proposed dynamic VM consolidation technique optimizes energy utilization while meeting strict SLA requirements and enhancing QoS metrics in CDCs. Extensive analyses were conducted using the Cloudsim toolkit to validate the approach's effectiveness. These analyses encompassed conditions such as random workloads in heterogeneous environments. The simulation results showed that GE‐EBWO outperforms other techniques and improves energy efficiency by 12%–15%. In addition, it significantly decreases VM migrations by 11%–14% compared to other advanced methods. The study validates the practicality of the proposed technique in moving towards environmentally friendly CDCs.

Service Selection under Uncertainty

Cloud computing has emerged as a popular computing paradigm, providing scalable and pay-per-use services to execute a variety of tasks. However, selecting the most suitable cloud services for a client can be challenging, as it involves taking into account the service characteristics and a set of client restrictions. In this paper, we formalize the problem of Service Selection under Uncertainty (SSuU) as an optimization problem. Our goal is to allocate tasks to appropriate cloud services while considering the probability of Service Level Agreement (SLA) violations and a set of client restrictions. We introduce an efficient dynamic programming approach to calculate the probability of SLA violations incrementally over time. We prove that the SSuU problem is strongly NP-complete, and propose an Integer Linear Programming (ILP) formulation and an Iterated Local Search-based algorithm for tackling it. To facilitate evaluation, we also introduce a set of benchmark instances for the SSuU problem. We extensively evaluate our proposed solutions on 94 input instances and compare them to the exact method (i.e., which always produces the optimal solution). Our results demonstrate that the metaheuristic approach is significantly faster and leads to exact solutions in the great majority of the evaluation scenarios.

Efficient resource allocation in cloud computing environments using AI-driven predictive analytics

This paper proposes an innovative AI-driven approach for efficient resource allocation in cloud computing environments using predictive analytics. The study addresses the critical challenge of optimizing resource utilization while maintaining high quality of service in dynamic cloud infrastructures. A hybrid predictive model combining XGBoost and LSTM networks is developed to forecast workload patterns across various time horizons. The model leverages historical data from a large-scale cloud environment, encompassing 1000 servers and over 52 million data points. A dynamic resource scaling algorithm is introduced, which integrates the predictive model outputs with real-time system state information to make proactive allocation decisions. The proposed framework incorporates advanced techniques such as workload consolidation, resource oversubscription, and elastic resource pools to maximize utilization efficiency. Experimental results demonstrate significant improvements in key performance indicators, including increasing resource utilization from 65% to 83%, reducing SLA violation rates from 2.5% to 0.8%, and enhancing energy efficiency, with PUE improving from 1.4 to 1.18. Comparative analysis shows that the proposed model outperforms existing prediction accuracy and resource allocation efficiency methods. The study contributes to the field by presenting a comprehensive, AI-driven solution that addresses the complexities of modern cloud environments and paves the way for more intelligent and autonomous cloud resource management systems.

Network load balancing and data categorization in cloud computing

Cloud computing (CC) is rising quickly as a successful model presenting an on-demand structure. In the CC, the present investigation shows that load-balancing methods established on meta-heuristics offer better solutions for appropriate scheduling and allotment of resources. Conversely, several traditional approaches believe in only some quality of service (QoS) metrics and reject several significant components. Network load balancing and data categorization (NBDC) is proposed. This approach aims to enhance load balancing in the cloud field. This approach consists of two phases: the support vector machine (SVM) algorithm-based data categorization and the ant colony optimization (ACO) algorithm for distributing the network load on the virtual machine (VM). The SVM algorithm performs several data formats, such as text, image, audio, and video, resultant data class that offers high categorization accuracy in the cloud. The ACO algorithm reaches an efficient load balancing based on the time of execution (TE), time of throughput (TT), time of overhead (TO), time of optimization, and migration count (MC). Simulation results related to the baseline approach demonstrate an enhanced system function in terms of service level agreement violation, throughput, execution time, energy utilization, and execution time.

The analysis of the models performance indicators by cloud and post-cloud computing paradigms

The work describes cloud and post-cloud computing paradigms, on the basis of which it is possible to model and build intelligent video surveillance systems. Based on the analysis of literary sources, it was concluded that performance indicators largely depend on the paradigm of the computational model, but the degree of importance of each performance indicator is different for different paradigms. Taking into account the peculiarities of different paradigms, the performance indicators characteristic of each of them are described. It has been found that the common indicators for evaluating all models developed on the basis of cloud and post-cloud computing paradigms (with varying degrees of importance) are: Throughput, Network Congestion, SLA Violation, Fault Tolerance, Statistical Analysis Measurements, The Number of Orchestration Decisions, The number of Contradictory Decisions, Time to Adaptation, Competition Ratio, Cache Hit Ratio, The number of Contradictory Actions, Oscillation Mitigation, Cost, Technique’s Overhead, Privacy. Based on the analysis of cloud and post-cloud computing simulators, it is established that the metrics of resource usage, energy, response time, and latency are those that can be used by researchers when conducting a comparative analysis of models built on the basis of the following paradigms: fog, edge, fog, and cloud computing. The practical significance of the conducted research is that, depending on the tasks that will be set before the future video surveillance system, researchers will be able to evaluate the effectiveness of using different architectures (Cloud, FOG, Mist, etc.) with the help of certain simulators. In the future, it is advisable to conduct a study of the degree of importance of common metrics for evaluating all models developed on the basis of cloud and post-cloud computing paradigms, and conduct their normalization for the possibility of conducting an adequate comparative analysis. It is also relevant in future research that it is advisable to conduct research using different types of sensor devices (surveillance cameras with different technical characteristics, unmanned aerial vehicles, etc.).

Efficient resource allocation in cloud environment using SHO-ANN-based hybrid approach

The cloud computing paradigm provides services to users in an on-demand fashion using high-speed Internet. This Internet-based computing paradigm provides resources on a rent basis without any fault. Virtual machine resource allocation is one of the challenging concerns in a cloud computing environment. The existing static, dynamic, and Meta-Heuristic approaches provide the solution to the virtual machine allocation problem. These techniques stuck with the local optimal solution. The slow convergence rate leads to the optimal solution locally and fails to provide the optimal solution Globally. This manuscript proposes a hybrid Spotted Hyena optimizer and artificial neural network, named the SHO-ANN technique, to provide a solution to the virtual machine assignment problem. The presented hybrid technique is evaluated and analyzed using performance metrics “Energy Consumption (Kwh) (8.54%), Host Utilization (24.8%), Average Execution Time(ms) (26.33%), SLA Violations (1.33%), and Number of Migrations (Counts) (19.73%)”. The spotted hyena optimizer is used to provide the vast data set to the ANN model for better accuracy. The hybrid approach provides an optimal solution globally with high convergence. The experimental results exhibit that the SHO-ANN outperforms the IqMc, SHO, and Genetic approaches using real workload scenarios and fabricated scenarios.

Revolutionizing Cloud Resource Allocation: Harnessing Layer-Optimized Long Short-Term Memory for Energy-Efficient Predictive Resource Management

INTRODUCTION: This is the introductory text. Accurate data center resource projection will be challenging due to the dynamic and constantly changing workloads of multi-tenant co-hosted applications. Resource Management in the Cloud (RMC) becomes a significant research component. In the cloud's easy service option, users can choose to pay a fixed sum or based on the amount of time. OBJECTIVES: The main goal of this study is systematic method for estimating future cloud resource requirements based on historical consumption. Resource distribution to users, who require a variety of resources, is one of cloud computing main objective in this study. METHODS: This article suggests a Layer optimized based Long Short-Term Memory (LOLSTM) to estimate the resource requirements for upcoming time slots. This model also detects SLA violations when the QoS value exceeds the dynamic threshold value, and it then proposes the proper countermeasures based on the risk involved with the violation. RESULTS: Results indicate that in terms of training and validation the accuracy is 97.6%, 95.9% respectively, RMSE and MAD shows error rate 0.127 and 0.107, The proposed method has a minimal training and validation loss at epoch 100 are 0.6092 and 0.5828, respectively. So, the suggested technique performed better than the current techniques. CONCLUSION: In this work, the resource requirements for future time slots are predicted using LOLSTM technique. It regularizes the weights of the network and avoids overfitting. In addition, the proposed work also takes necessary actions if the SLA violation is recognized by the model. Overall, the proposed work in this study shows better performance compared to the existing methods.

Efficient Resource Management in Cloud Environments: A Modified Feeding Birds Algorithm for VM Consolidation

Cloud data centers play a vital role in modern computing infrastructure, offering scalable resources for diverse applications. However, managing costs and resources efficiently in these centers has become a crucial concern due to the exponential growth of cloud computing. User applications exhibit complex behavior, leading to fluctuations in system performance and increased power usage. To tackle these obstacles, we introduce the Modified Feeding Birds Algorithm (ModAFBA) as an innovative solution for virtual machine (VM) consolidation in cloud environments. The primary objective is to enhance resource management and operational efficiency in cloud data centers. ModAFBA incorporates adaptive position update rules and strategies specifically designed to minimize VM migrations, addressing the unique challenges of VM consolidation. The experimental findings demonstrated substantial improvements in key performance metrics. Specifically, the ModAFBA method exhibited significant enhancements in energy usage, SLA compliance, and the number of VM migrations compared to benchmark algorithms such as TOPSIS, SVMP, and PVMP methods. Notably, the ModAFBA method achieved reductions in energy usage of 49.16%, 55.76%, and 65.13% compared to the TOPSIS, SVMP, and PVMP methods, respectively. Moreover, the ModAFBA method resulted in decreases of around 83.80%, 22.65%, and 89.82% in the quantity of VM migrations in contrast to the aforementioned benchmark techniques. The results demonstrate that ModAFBA outperforms these benchmarks by significantly reducing energy consumption, operational costs, and SLA violations. These findings highlight the effectiveness of ModAFBA in optimizing VM placement and consolidation, offering a robust and scalable approach to improving the performance and sustainability of cloud data centers.

E2SVM: Electricity-Efficient SLA-aware Virtual Machine Consolidation approach in cloud data centers.

Cloud data centers present a challenge to environmental sustainability because of their significant energy consumption. Additionally, performance degradation resulting from energy management solutions, such as virtual machine (VM) consolidation, impacts service level agreements (SLAs) between cloud service providers and users. Thus, to achieve a balance between efficient energy consumption and avoiding SLA violations, we propose a novel VM consolidation algorithm. Conventional algorithms result in unnecessary migrations when improperly selecting VMs. Therefore, our proposed E2SVM algorithm addresses this issue by selecting VMs with high load fluctuations and minimal resource usage from overloaded servers. These selected VMs are then placed on normally loaded servers, considering their stability index. Moreover, our approach prevents server underutilization by either applying all or no VM migrations. Simulation results demonstrate a 12.9% decrease in maximum energy consumption compared with the minimum migration time VM selection policy. In addition, a 47% reduction in SLA violations was observed when using the medium absolute deviation as the overload detection policy. Therefore, this approach holds promise for real-world data centers because it minimizes energy waste and maintains low SLA violations.

An integrated optimization method to task scheduling and VM placement for green datacenters

In the realm of cloud computing, effective resource allocation can significantly enhance the energy efficiency of datacenters. Task scheduling and Virtual Machine Placement (VMP) are two pivotal aspects of resource allocation. However, in current research, they are often treated separately, overlooking the potential for integrated optimization. In this paper, we propose an integrated solution for task scheduling and VMP in energy-efficient datacenters, based on queueing theory and Deep Reinforcement Learning (DRL) methods. This novel and comprehensive approach provides an alternative perspective for resource scheduling strategies in datacenters. We construct a queueing theory model for task scheduling, aiming to minimize the number of VMs that need to be instantiated, while ensuring that Service Level Agreement (SLA) violation remains at a low level. Furthermore, we design a VMP algorithm based on DRL for real-time selection of Physical Hosts (PHs) for deploying VMs. Finally, we conduct a simulation evaluation using a small-scale datacenter. The experimental results demonstrate that our method consistently ensures a lower rate of SLA violation. Compared to existing algorithms, the DRL-based VMP algorithm enables a more balanced utilization of the various resources in the PHs and reduces the total power consumption of the datacenter by more than 10% on average.

UAVs-assisted QoS guarantee scheme of IoT applications for reliable mobile edge computing

Unmanned aerial vehicles (UAVs) assisted mobile edge computing (MEC) is an emerging network architecture that has been considered as a promising transformative service paradigm. It enhances the coverage of intelligent mobile communication network and promotes the rapid development of a wide range of internet of things (IoT) applications. However, the ever-increasing data transmission demands given rise by these IoT applications based on MEC have posed significant challenges to the service of quality (QoS) guarantee of UAVs-assisted MEC (UAVs-MEC) network. To address the issue, we propose a novel UAVs-MEC QoS guarantee scheme that can enhances the reliability of IoT applications in UAVs-MEC network. Specifically, a dynamic QoS-aware based service level agreement (SLA) compliance verification model for MEC service is proposed in the scheme. It can monitor and detect the consistency of dynamic QoS of IoT applications through UAVs to ensure the continuous SLA compliance of service provided by MEC. Additionally, an objective weight assignment method and an adaptive update method are presented to improve the accuracy and scalability of the scheme. Finally, the simulation experiments conducted using a real-world dataset show that the proposed scheme can efficiently and effectively verify the SLA compliance of MEC service for guaranteeing QoS of IoT applications in UAVs-MEC context while outperforms other existing SLA violation detection methods.

An efficient dynamic migration and consolidation method of VMs based on improved K-nearest neighbor algorithm and attribute weighting

Virtual machine (VM) consolidation and migration that only consider current workload can result in excessive unnecessary migrations. To address this issue, a VM consolidation algorithm based on resource utilization prediction is proposed. An improved K-nearest neighbor (KNN) classification algorithm weighted by attribute inconsistency is proposed to predict the workload of both the host and the VMs. Firstly, two distributions are partitioned according to the neighboring relationship for comparing consistency. Then, an inconsistency evaluation function based on earth mover’s distance (EMD) is designed to measure the inconsistency between the neighboring sample set of each sample under each attribute and the equivalent partition refined by the decision attribute. Finally, the inconsistency level of the neighboring samples is transformed into the importance of the corresponding attribute to implement the attribute weighting KNN classifier. When selecting the source host and target host for VM migration, both current and predicted overloads are considered to avoid unnecessary VM migrations. Simulation tests were performed with random and realistic workloads, and the results show that the proposed method can reduce the overall energy consumption of the host, while also reducing service level agreement (SLA) violations and VM migration.

A workload prediction model for reducing service level agreement violations in cloud data centers

Cloud computing has become an emerging technology that offers services based on the pay-as-usage model. The cloud provides several advantages, but these advantages come with challenges, such as reducing Service Level Agreement (SLA) violations, efficient resource utilization, reducing energy consumption, etc., needing attention to leverage customer satisfaction and benefit cloud service providers. Workload prediction is a strategy that provides many benefits: reduced SLA violation, resource scaling, and resource optimization by predicting future workload. However, due to the varying workload of cloud applications, it is difficult to predict the workload accurately, and it fails for long-term dependencies. We propose a methodology based on Multiplicative Long Short Term Memory (mLSTM) that allows input-dependent transitions and considers long-term dependencies to predict the workload to address this issue. The proposed method is implemented and compared with other variants of LSTM used in literature for workload prediction purposes. The proposed work outperforms existing variants of LSTM in terms of prediction accuracy.

The drift handling framework for open radio access networks: An experimental evaluation

The Open Radio Access Network (i.e., Open RAN) aims to transform the inflexible, proprietary RAN into one that is flexible, programmable, and disaggregated to provide intelligent solutions at the network edge for delivering the highly dynamic service demands of Beyond 5G (B5G) networks. Open RAN Working Group-2 (WG2) focuses on the architecture and standards of Artificial Intelligence/Machine Learning (AI/ML) to meet the Service Level Agreement (SLA) requirements for emerging use cases. The inherent dynamic nature of B5G networks and their applications often leads to AI/ML model performance degradation (i.e., drift), resulting in violations of SLAs, over- or under-provisioning of resources, etc.This paper proposes a drift handling framework which includes drift detection, analysis, and its adaptation to minimize the SLA violations and efficient resource utilization. The proposed drift handling framework is evaluated across three diverse use cases: (i) Quality of Service prediction, (ii) Channel Quality Indicator prediction, and (iii) User traffic prediction. The experimental evaluations are performed by using the Open RAN Software Community, and OpenAirInterface FlexRIC platforms with a real-time dataset. Results demonstrate that the proposed drift handling framework outperforms all other considered state-of-the art frameworks for all the scenarios in terms of the drift detection time, accuracy, precision, F1-score, etc.

Hybrid Elastic Scaling Strategy for Container Cloud based on Load Prediction and Reinforcement Learning

To harness the advantages of both proactive and responsive scaling, adapting to various workload scenarios, this paper introduces a container hybrid scaling strategy called HyPredRL, rooted in load prediction and reinforcement learning. Within the proactive scaling module RL-PM, a load prediction model, MSC-LSTM, predict workloads and, in conjunction with current workload states, leverages reinforcement learning agents for intelligent scaling decisions. The responsive scaling strategy, SLA-HPA, enhances Kubernetes’ native scaling strategy, which primarily considers resource utilization, by incorporating response time metrics. Ultimately, a hybrid scaling controller is designed, applying the principles of “rapid scaling out” and “balanced conflicts” to coordinate proactive and responsive scaling. Experimental results demonstrate that HyPredRL outperforms existing methods in SLA violation rate, resource utilization, and request response time, effectively improving application performance and scalability.

Optimizing Resource Management for Shared Microservices: A Scalable System Design

A common approach to improving resource utilization in data centers is to adaptively provision resources based on the actual workload. One fundamental challenge of doing this in microservice management frameworks, however, is that different components of a service can exhibit significant differences in their impact on end-to-end performance. To make resource management more challenging, a single microservice can be shared by multiple online services that have diverse workload patterns and SLA requirements. We present an efficient resource management system, namely Erms, for guaranteeing SLAs with high probability in shared microservice environments. Erms profiles microservice latency as a piece-wise linear function of the workload, resource usage, and interference. Based on this profiling, Erms builds resource scaling models to optimally determine latency targets for microservices with complex dependencies. Erms also designs new scheduling policies at shared microservices to further enhance resource efficiency. Experiments across microservice benchmarks as well as trace-driven simulations demonstrate that Erms can reduce SLA violation probability by 5× and more importantly, lead to a reduction in resource usage by 1.6×, compared to state-of-the-art approaches.