Testing Cloud Benchmark Scalability with Cassandra

Abramova, Veronika; Bernardino, Jorge; Furtado, Pedro

doi:10.1109/services.2014.81

Cited by 21 publications

(15 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…However, unlike these and other works (e.g. [6,7,8,9,10]) studying and comparing the maximum databases throughput we put the focus on analysing the dynamic aspects of the Cassandra performance under different consistency settings. In particular, we analyse how the database latency and throughput depend on a current workload (i.e.…”

Section: Cassandra Deployment Setupmentioning

confidence: 99%

“…where , probabilities of read/write requests, P Read + P Write = 1. Table IV provides some estimates of the Cassandra latency for different settings guaranteeing the strong consistency under a mixed read/write workload using (2-7) and (8)(9)(10). The Cassandra database is known for extremely fast writes/updates.…”

Section: ) 'Read One -Write All' (1r-3w); 2) 'Read Quorum -Write Quomentioning

confidence: 99%

“…There have been a number of studies, e.g. [6,7,8,9,10], evaluating and comparing the performance of different NoSQL databases.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Interplaying Cassandra NoSQL Consistency and Performance: A Benchmarking Approach

Gorbenko

Romanovsky

Tarasyuk

2020

Communications in Computer and Information Science

View full text Add to dashboard Cite

This experience report analyses performance of the Cassandra NoSQL database and studies the fundamental trade-off between data consistency and delays in distributed data storages. The primary focus is on investigating the interplay between the Cassandra performance (response time) and its consistency settings. The paper reports the results of the read and write performance benchmarking for a replicated Cassandra cluster, deployed in the Amazon EC2 Cloud. We present quantitative results showing how different consistency settings affect the Cassandra performance under different workloads. One of our main findings is that it is possible to minimize Cassandra delays and still guarantee the strong data consistency by optimal coordination of consistency settings for both read and write requests. Our experiments show that (i) strong consistency costs up to 25% of performance and (ii) the best setting for strong consistency depends on the ratio of read and write operations. Finally, we generalize our experience by proposing a benchmarking-based methodology for run-time optimization of consistency settings to achieve the maximum Cassandra performance and still guarantee the strong data consistency under mixed workloads.

show abstract

Section: Cassandra Deployment Setupmentioning

confidence: 99%

Section: ) 'Read One -Write All' (1r-3w); 2) 'Read Quorum -Write Quomentioning

confidence: 99%

See 1 more Smart Citation

Interplaying Cassandra NoSQL Consistency and Performance: A Benchmarking Approach

Gorbenko

Romanovsky

Tarasyuk

2020

Communications in Computer and Information Science

View full text Add to dashboard Cite

show abstract

“…Even if relational database-management system (RDBMS) stores its data in memory, in general the performance is worse than that of NoSQL databases. Abramova et al [14] tested the performance of Cassandra based on a number of factors, including the number of nodes, workload characteristics, number of threads, and data size, and analyzed whether it provides the desired acceleration and scalability attributes. Scaling nodes and the number of data-sets do not guarantee performance.…”

Section: Related Workmentioning

confidence: 99%

Evaluating Riak Key Value Cluster for Big Data

Rmis

Topcu

2020

Teh. vjesn.

View full text Add to dashboard Cite

NoSQL database has become an important alternative to traditional relational databases. Those databases are prepared by the management of large, continuously and variably changing data sets. They are widely used in cloud databases and distributed systems. With NoSQL databases, static schemes and many other restrictions are avoided. In the era of big data, such databases provide scalable high availability solutions. Their key-value feature allows fast retrieval of data and the ability to store a lot of it. There are many kinds of NoSQL databases with various performances. Therefore, comparing those different types of databases in terms of performance and verifying the relationship between performance and database type has become very important. In this paper, we test and evaluate the Riak key-value database for big data clusters using benchmark tools, where huge amounts of data are stored and retrieved in different sizes in a distributed database environment. Execution times of the NoSQL database over different types of workloads and different sizes of data are compared. The results show that the Riak key-value is stable in execution time for both small and large amounts of data, and the throughput performance increases as the number of threads increases.

show abstract

“…In [14], the authors analyze the scalability, using the Yahoo Cloud Serving Benchmark with Cassandra. The mOSAIC Benchmarking Framework [15] for the mOSAIC platform ).…”

Section: Related Workmentioning

confidence: 99%

Towards a unified customer aware figure of merit for CSP selection

Zant

Gagnaire

2015

J Cloud Comp

View full text Add to dashboard Cite

The current cloud computing market lacks of clear comparison between the Cloud service providers (CSPs) offerings. This is due to the heterogeneity in the virtual machines (VMs) configurations and their prices which differ among the CSPs. Big players in the market offer different configurations of fixed size VMs. Cloud customers have to choose the CSP that best fits their requirements. In the actual market, and with the limited performance information provided by the CSPs to the cloud users, the choice of the CSP can be a problem for the customers. In our paper, and in the context of the Easi-Clouds (project I, Easi Clouds. http://www.easi-clouds.eu/) a European ITEA 2 research project, we propose a set of performance tests based on real measurements to classify the CSPs based on their performance score as well as their proposed price. We used a set of benchmarks to test the performances of four VMs' sizes (Small (S), Medium (M), Large (L), and Xlarge (XL)) from each one of the biggest eight CSPs (Amazon, Softlayer, Rackspace, Google, Microsoft Azure, Aruba, Digital Ocean, Joyent). We try to compare the performance based on seven different metrics (CPU performance, Memory performance, Disk I/O performance, Mean Response time (MRT), Provisioning time, Availability, and Variability). In a second step, we include the price to have a performance vs. price value figure. In a final step, we propose a new method that let the user specify the importance of each performance's metric as well as the importance of the price to classify the CSPs based on the criterions of the customers. We come up with a unified customer aware figure of merit helping the cloud customers to select the most suitable CSP based on their own requirements.

show abstract

Testing Cloud Benchmark Scalability with Cassandra

Cited by 21 publications

References 15 publications

Interplaying Cassandra NoSQL Consistency and Performance: A Benchmarking Approach

Interplaying Cassandra NoSQL Consistency and Performance: A Benchmarking Approach

Evaluating Riak Key Value Cluster for Big Data

Towards a unified customer aware figure of merit for CSP selection

Contact Info

Product

Resources

About