Scaling the monitoring of approximate top-k queries in streaming windows

Rampérez, Víctor; Zahmatkesh, Shima; Valle, Emanuele Della

doi:10.1109/bigdata52589.2021.9672041

Cited by 2 publications

(3 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As mentioned above, the method for predicting the SLA parameters of the system are different in each predictive autoscaling system [11,12,14], despite this, the use of a monitor component is essential to measure the chosen prediction parameter(s), which, in our case, are the window processing time (WPT) [23], the response time (RT), and the throughput (TH). Together, these measurements represent a high-fidelity snapshot of the system status.…”

Section: Prediction Of Sla Parametersmentioning

confidence: 99%

“…e system we have used for the experiments, a complex event processing (CEP) distributed system, that calculates the topk using the Min-TopK + N algorithm developed in [23] has been deployed using Apache Kafka and Zoopeker. For these experiments, we have used two T2 EC2 instances of Amazon Web Services cloud, each with 16 GB of RAM, 4 vCPUs running the 4.14 Amazon Linux kernel and Java OpenJDK 11.0.7 2020-04-14 LTS.…”

Section: Experimental Environmentmentioning

confidence: 99%

“…We increased the number of users to 200 k, randomized the number of followers of each user, and added more user interactions to include some saturation points while keeping the load true to the original. For more information, see [23]. 6 and 7 show the effects of a workload on the SLA parameters, in this case, the window processing time.…”

Section: Experimental Environmentmentioning

confidence: 99%

See 2 more Smart Citations

Towards SLA-Driven Autoscaling of Cloud Distributed Services for Mobile Communications

Miguel

Rampérez

Soriano

et al. 2022

Mobile Information Systems

View full text Add to dashboard Cite

In recent years cloud computing has established itself as the computing paradigm that supports most distributed systems, which are essential in mobile communications, such as publish-subscribe (pub/sub) systems or complex event processing (CEP). The cornerstone of cloud computing is elasticity, and today’s autoscaling systems leverage that property by making scaling decisions based on estimates of future workload to satisfy service level agreements (SLAs). However, these autoscaling systems are not generic enough, as the workload definition is application-based. On the other hand, the workload prediction needs to be mapped in terms of SLA parameters, which introduces a double prediction problem. This work presents an empirical study on the relationship between different types of workloads in the literature and their relationship in terms of SLA parameters in the context of mobile communications. In addition, more than 30 prediction models have been trained using different techniques (time series analysis, regression, random forests) to test which ones offer better prediction results of the SLA parameters based on the type of workload and the prediction horizon. Finally, a series of conclusions on the predictive models to be used as a first step towards an autonomous decision system are presented.

show abstract