Micro-Service Placement Policies for Cost Optimization in Kubernetes

Aznavouridis, Alkiviadis; Tsakos, Konstantinos; Petrakis, Euripides G. M.

doi:10.1007/978-3-030-99587-4_35

Cited by 9 publications

(13 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The graduated project maturity level shows that Prometheus is stable for production and has great potential to integrate with modern orchestrators such as Kubernetes. Prometheus has also become the monitoring solution for many research works [27]- [32].…”

Section: Motivationmentioning

confidence: 99%

Acala: Aggregate Monitoring for Geo-Distributed Cluster Federations

Huang

Pierre

2023

Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing

View full text Add to dashboard Cite

Monitoring plays a vital role in geo-distributed cluster federation environments to accurately schedule applications across geographically dispersed computing resources. However, using a fixed frequency for collecting monitoring data from clusters may waste network bandwidth and is not necessary for ensuring accurate scheduling. In this paper, we propose Adaptive Prometheus Federation (AdapPF), an extension of the widelyused open-source monitoring tool, Prometheus, and its feature, Prometheus Federation. AdapPF aims to dynamically adjust the collection frequency of monitoring data for each cluster in geodistributed cluster federations. Based on actual deployment in the geo-distributed Grid'5000 testbed, our evaluations demonstrate that AdapPF can achieve comparable results to Prometheus Federation with 5-seconds scrape interval while reducing crosscluster network traffic by 36%.

show abstract

Section: Motivationmentioning

confidence: 99%

Acala: Aggregate Monitoring for Geo-Distributed Cluster Federations

Huang

Pierre

2023

Proceedings of the 38th ACM/SIGAPP Symposium on Applied Computing

View full text Add to dashboard Cite

show abstract

“…Additionally, arranging the components in a way that maintains system attributes such as availability and low latency introduces intricacies [3,4]. The challenges of debugging [5][6][7][8][9] and component arrangement [10] have been addressed through the implementation of distributed tracing techniques [11][12][13][14][15][16][17][18].…”

Section: Introductionmentioning

confidence: 99%

Vnode: Low-Overhead Transparent Tracing of Node.js-Based Microservice Architectures

Kabamba,

Khouzam,

Dagenais

2023

Future Internet

View full text Add to dashboard Cite

Tracing serves as a key method for evaluating the performance of microservices-based architectures, which are renowned for their scalability, resource efficiency, and high availability. Despite their advantages, these architectures often pose unique debugging challenges that necessitate trade-offs, including the burden of instrumentation overhead. With Node.js emerging as a leading development environment recognized for its rapidly growing ecosystem, there is a pressing need for innovative performance debugging approaches that reduce the telemetry data collection efforts and the overhead incurred by the environment’s instrumentation. In response, we introduce a new approach designed for transparent tracing and performance debugging of microservices in cloud settings. This approach is centered around our newly developed Internal Transparent Tracing and Context Reconstruction (ITTCR) technique. ITTCR is adept at correlating internal metrics from various distributed trace files to reconstruct the intricate execution contexts of microservices operating in a Node.js environment. Our method achieves transparency by directly instrumenting the Node.js virtual machine, enabling the collection and analysis of trace events in a transparent manner. This process facilitates the creation of visualization tools, enhancing the understanding and analysis of microservice performance in cloud environments. Compared to other methods, our approach incurs an overhead of approximately 5% on the system for the trace collection infrastructure while exhibiting minimal utilization of system resources during analysis execution. Experiments demonstrate that our technique scales well with very large trace files containing huge numbers of events and performs analyses in very acceptable timeframes.

show abstract

“…In our recent work [5], all these objectives are met together by grouping the application microservices into smaller groups with high-affinity rates and by placing each such group within the same node. The problem of microservices placement is formulated as graph clustering (or graph partitioning).…”

Section: Introductionmentioning

confidence: 99%

“…Placing these services in the same node reduces the traffic exchanged between the infrastructure's nodes (i.e., egress traffic). In [5], several graph partitioning algorithms are tested and their placement solutions are compared against the solution of the default Kubernetes Scheduler in the Google Kubernetes Engine (GKE). Our proposed Bisecting K-Means (BKM) and Heuristic First Fit (HFF) [6] proved to be more efficient by optimizing almost all cost factors (i.e., number of nodes, egress traffic, infrastructure hosting cost).…”

Section: Introductionmentioning

confidence: 99%

“…Application deployment comes along with appropriate supporting services that produce the graph of an application, a service mesh [10], with monitoring services and a middleware service that automates service placement to GKE (i.e., translates the output of graph clustering to Docker Compose). We present comparisons of the proposed fuzzy method with placement with Bisecting K-Means [5], Heuristic First Fit [6], and the default Kubernetes Scheduler [1] based on several performance metrics (i.e., response time, infrastructure resources, and hosting cost). For proof of concept, service placement results from all methods are reported for two real-world applications (i.e., the Google Online Boutique eShop demo application [11] and iXen IoT architecture [12].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

ModSoft-HP: Fuzzy Microservices Placement in Kubernetes

Petrakis,

Skevakis,

Eliades

et al. 2023

Electronics

Self Cite

View full text Add to dashboard Cite

The growing popularity of microservices architectures generated the need for tools that orchestrate their deployment in containerized infrastructures, such as Kubernetes. Microservices running in separate containers are packed in pods and placed in virtual machines (nodes). For applications with multiple communicating microservices, the decision of which services should be placed in the same node has a certain impact on both the running time and the operation cost of an application. The default Kubernetes scheduler is not optimal in that case. In this work, the service placement problem is treated as graph clustering. An application is modeled using a graph with nodes and edges representing communicating microservices. Graph clustering partitions the graph into clusters of microservices with high-affinity rates. Then, the microservices of each cluster are placed in the same Kubernetes node. A class of methods resorts to hard clustering (i.e., each microservice is placed in exactly one node). We advocate that graph clustering should be fuzzy to allow high-utilized microservices to run in more than one instance (i.e., pods) in different nodes. ModSoft-HP Scheduler is a custom Kubernetes scheduler that takes scheduling decisions based on the results of the ModSoft fuzzy clustering method followed by heuristic packing (HP). For proof of concept, the workloads of two applications (i.e., an e-commerce application, eShop, and an IoT architecture) are given as input to the default Kubernetes Scheduler, the Bisecting K-means, and the Heuristic First Fit (hard) clustering schedulers and to the ModSoft-HP fuzzy clustering method. The experimental results demonstrate that ModSoft-HP can achieve up to 90% reduction of egress traffic, up to 20% savings in response time, and up to 25% less hosting costs compared to service placement with the default Kubernetes Scheduler in the Google Kubernetes Engine.

show abstract

Micro-Service Placement Policies for Cost Optimization in Kubernetes

Cited by 9 publications

References 16 publications

Acala: Aggregate Monitoring for Geo-Distributed Cluster Federations

Acala: Aggregate Monitoring for Geo-Distributed Cluster Federations

Vnode: Low-Overhead Transparent Tracing of Node.js-Based Microservice Architectures

ModSoft-HP: Fuzzy Microservices Placement in Kubernetes

Contact Info

Product

Resources

About