A5071626310- Cloud Computing and Resource Management
- Software System Performance and Reliability
- IoT and Edge/Fog Computing
- Distributed and Parallel Computing Systems
- Parallel Computing and Optimization Techniques
- Distributed systems and fault tolerance
- Caching and Content Delivery
- Advanced Data Storage Technologies
- Advanced Queuing Theory Analysis
- Software-Defined Networks and 5G
- Age of Information Optimization
- Real-Time Systems Scheduling
- Advanced Software Engineering Methodologies
- Formal Methods in Verification
- Blockchain Technology Applications and Security
- Numerical Methods and Algorithms
- Business Process Modeling and Analysis
- Embedded Systems Design Techniques
- Network Traffic and Congestion Control
- Green IT and Sustainability
- Privacy-Preserving Technologies in Data
- Cryptographic Implementations and Security
- Access Control and Trust
- Cellular Automata and Applications
- Electromagnetic Simulation and Numerical Methods
Umeå University
2015-2024
Institut national de recherche en informatique et en automatique
2011-2021
Politecnico di Milano
2021
Ericsson (Sweden)
2021
University of California, Santa Barbara
2021
William & Mary
2021
IBM Research (China)
2021
Université de Rennes
2021
Intel (Germany)
2021
Sesame Workshop
2021
Self-adaptation is a first class concern for cloud applications, which should be able to withstand diverse runtime changes. Variations are simultaneously happening both at the infrastructure level - example hardware failures and user workload flash crowds. However, robustly withstanding extreme variability, requires costly over-provisioning.
The pervasiveness and growing complexity of software systems are challenging engineering to design that can adapt their behavior withstand unpredictable, uncertain, continuously changing execution environments. Control theoretical adaptation mechanisms have received interest from the community in last few years for mathematical grounding, allowing formal guarantees on controlled systems. However, most these tailored specific applications hardly be generalized into broadly applicable...
The software engineering community has proposed numerous approaches for making self-adaptive. These take inspiration from machine learning and control theory, constructing that monitors modifies its own behavior to meet goals. Control in particular, received considerable attention as it represents a general methodology creating adaptive systems. Control-theoretical implementations, however, tend be ad hoc. While such solutions often work practice, is difficult understand reason about the...
Custom operators, working at custom precisions, are a key ingredient to fully exploit the FPGA flexibility advantage for high-performance computing. Unfortunately, such operators costly design, and application designers tend rely on less efficient off-the-shelf operators. To address this issue, an open-source architecture generator framework is introduced. Its salient features easy learning curve from VHDL, ability embed arbitrary synthesizable VHDL code, portability mainstream targets...
Resource provisioning in cloud computing is typically coarse-grained. For example, entire CPU cores may be allocated for periods of up to an hour. The Resource-as-a-Service concept has been introduced improve the efficiency resource utilization clouds. In this concept, resources are terms core fractions, with granularities seconds. Such infrastructures could created using existing technologies such as lightweight virtualization LXC or by exploiting Xen hyper visor's capacity vertical...
Summary This experience article describes lessons learned as we conducted experiments in a Kubernetes‐based environment, the most notable of which was that performance both Kubernetes control plane and deployed application depends strongly unexpected ways on etcd database. The contains (a) detailed descriptions how networking with without Istio works Kubernetes, based Flannel Container Networking Interface (CNI) provider VXLAN mode IP Virtual Server (IPVS)‐backed Services, (b) comprehensive...
Vertical elasticity is recognized as a key enabler for efficient resource utilization of cloud infrastructure through fine-grained provisioning, e.g., allowing CPU cycles to be leased short few seconds. However, little research has been done support vertical where the focus mostly on single resource, either or memory, while an application may need arbitrary combinations these resources at different stages its execution. Nonetheless, existing techniques cannot readily used as-is without...
We focus on improving resilience of cloud services (e.g., e-commerce website), when correlated or cascading failures lead to computing capacity shortage. study how extend the classical service architecture composed a load-balancer and replicas with recently proposed self-adaptive paradigm called brownout. Such are able reduce their requirements by degrading user experience disabling recommendations). Combining brownout is date an open practical problem. The issue ensure that replica...
To reduce the congestion due to future bandwidth-hungry applications in domains such as Health care, Internet of Things (IoT), etc., we study benefit introducing additional Data Centers (DCs) closer network edge for optimal application placement. Our shows that layer DCs a Mobile Edge Network (MEN) infrastructure is cost beneficial having their strong demand locality and scenarios where large capacity deployed at DCs. The savings can go up 67%. Additional intermediate close root DC be...
A microservice architecture features hundreds or even thousands of small loosely coupled services with multiple instances. Because performance depends on many factors including the workload, inter-service traffic management is complex in such dynamic environments. Service meshes aim to handle this complexity and facilitate management, observability, communication between microservices. provide various policies as circuit breaking retry mechanisms, which are claimed protect microservices...
Mobile Edge Cloud (MEC) is a platform complementing traditional centralized clouds, consisting in moving computing and storage capacity closer to users - e.g., as Data Centers (EDC) base stations order reduce application-level latency network bandwidth. The bounded coverage radius of station limited each EDC intertwined with user mobility challenge the operator's ability perform adjustment planning. To face this challenge, proactive resource provisioning can be performed. usage estimated...
Mobile Edge Clouds (MECs) is a promising computing platform to overcome challenges for the success of bandwidth-hungry, latency-critical applications by distributing and storage capacity in edge network as Data Centers (EDCs) within close vicinity end-users. Due heterogeneous distributed resource EDCs, application deployment flexibility coupled with user mobility, MECs bring significant control allocation provisioning. In order develop self-managed system which efficiently decides how much...
Resource overbooking is an admission control technique to increase utilization in cloud environments. However, due uncertainty about future application workloads, may result overload situations and deteriorated performance. We mitigate this using brownout, a feedback approach performance steering, that ensures graceful degradation during load spikes thus avoids overload. Additionally, brownout management information included into the system, enabling development of improved reactive methods...
Applications hosted in the cloud have become indispensable several contexts, with their performance often being key to business operation and running costs needing be minimized. To minimize costs, most modern virtualization technologies such as Linux Containers, Xen, KVM offer powerful resource control primitives for individual provisioning - that enable adding or removing of fraction cores and/or megabytes memory short few seconds. Despite technology ready, there is a lack proper techniques...
Virtualization solutions based on hypervisors or containers are enabling technologies for scalable, flexible, and cost-effective resource sharing. As the fundamental limitations of each technology yet to be understood, they need regularly reevaluated better understand trade-off provided by latest technological advances. This paper presents an in-depth quantitative analysis virtualization overheads in these two groups systems their gaps relative native environments a diverse set workloads...
Cloud applications are often subject to unexpected events like flash crowds and hardware failures. Without a predictable behaviour, users may abandon an unresponsive application. This problem has been partially solved on two separate fronts: first, by adding self-adaptive feature called brownout inside cloud bound response times modulating user experience, and, second, introducing replicas - copies of the having same function-alities for redundancy load-balancer direct incoming traffic.
Due to fierce competition, cloud providers need run their data-centers efficiently. One of the issues is increase data-center utilization while maintaining applications' performance targets. Achieving high meeting difficult, as overload may lead poor hosted services. Service differentiation has been proposed control which services get degraded. However, current approaches are capacity-based, oblivious observed each service and cannot divide available capacity among so minimize overall...
Cloud Infrastructure as a Service (IaaS) providers continually seek higher resource utilization to better amortize capital costs. Higher not only can enable profit for IaaS but also provides mechanism raise energy efficiency; therefore creating greener cloud services. Unfortunately, achieving high is difficult mainly due infrastructure needing maintain spare capacity service demand fluctuations.
Mobile Edge Clouds (MECs) are distributed platforms in which distant data-centers complemented with computing and storage capacity located at the edge of network. Their wide resource distribution enables MECs to fulfill need low latency high bandwidth offer an improved user experience.
Site Reliability Engineers are at the center of two tensions: On one hand, they need to respond alerts within a short time, restore non-functional system. other response times is disruptive everyday life and lead alert fatigue. To alleviate this tension, many resource management mechanisms proposed handle overload mitigate faults. One recent such mechanism circuit breaking in service meshes. Circuit rejects incoming requests protect latency expense availability (successfully answered...
Microservice-based architectures have become ubiq-uitous in large-scale software systems. Experimental cloud re-searchers constantly propose enhanced resource management mechanisms for such These need to be eval-uated using both realistic and flexible microservice benchmarks study which ways diverse application characteristics can affect their performance scalability. However, current mi-croservice limitations including static compu-tational complexity, limited architectural scale, fixed...
The extreme adoption rate of container technologies along with raised security concerns have resulted in the development multiple alternative runtimes targeting through additional layers indirection. In an apples-to-apples comparison, we deploy three same Kubernetes cluster, focused Kata and gVisor, as well default runtime runC. Our evaluation based on real applications demonstrate that runC outperforms more secure alternatives up to 5x, gVisor deploys containers 2x faster than Kata, but...