Authors:
Eduardo Carvalho
1
;
Raul Barbosa
2
and
Jorge Bernardino
1
Affiliations:
1
Polytechnic of Coimbra, Portugal
;
2
University of Coimbra, Portugal
Keyword(s):
Cloud Benchmarking, Dependability, Fault Tolerance, Cloud Computing.
Abstract:
Current benchmarks focus on evaluating performance with little efforts made to evaluate the dependability characteristics of the cloud. Cloud computing has several advantages like scalability, elasticity and cost reduction, this led companies to move their applications to the cloud. The availability of applications and consequently businesses are then dependant on the cloud’s efforts to keep its services running. To guarantee reliability and trust, benchmarking dependability is a challenging task because of the cloud layered model which makes it difficult to predict the root of faults as higher layers are dependant of lower layers. By integrating dependability in benchmarks as a metric, we can evaluate how well does the cloud handle itself when faults occur, prevent those faults and check not only raw performance but also trust. In this paper, we study the following cloud benchmarks: Spec IaaS 2016, TPCx-V, YCSB, Perfkit Benchmarker, CloudBench, DS-Bench/D-Cloud, and evaluate if they
are suitable for benchmarking dependability.
(More)