Performance evaluation of containers and virtual machines when running Cassandra workload concurrently

Abstract

SummaryNoSQL distributed databases are often used as Big Data platforms. To provide efficient resource sharing and cost effectiveness, such distributed databases typically run concurrently on a virtualized infrastructure that could be implemented using hypervisor‐based virtualization or container‐based virtualization. Hypervisor‐based virtualization is a mature technology but imposes overhead on CPU, networking, and disk. Recently, by sharing the operating system resources and simplifying the deployment of applications, container‐based virtualization is getting more popular. This article presents a performance comparison between multiple instances of VMware VMs and Docker containers running concurrently. Our workload models a real‐world Big Data Apache Cassandra application from Ericsson. As a baseline, we evaluated the performance of Cassandra when running on the nonvirtualized physical infrastructure. Our study shows that Docker has lower overhead compared with VMware; the performance on the container‐based infrastructure was as good as on the nonvirtualized. Our performance evaluations also show that running multiple instances of a Cassandra database concurrently affected the performance of read and write operations differently; for both VMware and Docker, the maximum number of read operations was reduced when we ran several instances concurrently, whereas the maximum number of write operations increased when we ran instances concurrently.

Performance Assessment of Using Docker for Selected MPI Applications in a Parallel Environment Based on Commodity Hardware

“…A slightly similar performance investigation of multiple instances of VMWare VMs and Docker containers running concurrently, albeit in the context of multiple instances of a Cassandra database running concurrently was presented in paper [7]. Firstly, for one Cassandra cluster, the authors showed averaged latency of 1.35 ms for Docker vs. 1.9 ms VMWare and 1.2 non-virtualized for replication factor 1, 1.9 ms, 4.4 ms and 2.1 ms for replication factor 2 and 2.2 ms, 5.45 ms and 2.1 ms for replication factor 3 for Docker, VMWare and the non-virtualized environment, respectively.…”

“…We can note that, the best speed-up for DAC and Host is reached using 256 processes (oversubscription) although it is very close to 64 and 128 process configurations, for DAC and Docker it is best for 64 processes; for MasterSlave, using 64 processes; and SPMD-128 processes and falling afterwards. Speed-up was calculated using function (7). speedup(mach, param, prog, proc) = medianExecutionTime(mach, param, prog, 1) medianExecutionTime(mach, param, prog, proc) We shall note that we generally expect a decrease in execution time with an increasing number of processes assuming these run on separate physical or logical cores in the system, in our case 64 and 128 respectively, because such a scenario allows shortening the computational phase.…”

“…In the latter, it is of special interest to investigate potential performance impacts as HPC systems have been developed primarily with performance in mind and both execution times as well as scalability measured with speed-ups are of key importance. It can be noticed that virtualized environments can be used to run various workloads in parallel, from, e.g., multiple Cassandra clusters [7], through multiple DeviceHive IoT middleware containers [3] to compute intensive HPC applications [8].…”

CoreKube: An Efficient, Autoscaling and Resilient Mobile Core System

“…For orchestrating virtualized applications and services, containers are preferred over virtual machines (VMs) as they are lightweight [26,65] and have low startup times [19]. Kubernetes (K8s) [27] is the most popular open source container orchestration system [21].…”

An evaluation of relational and NoSQL distributed databases on a low-power cluster

“…In [ 21 ], the authors compared two orchestration tools and showed that Docker Swarm outperformed Kubernetes in resource usage and performance efficiency on a Raspberry Pi 3 cluster. In [ 22 ], they compared container-based Cassandra instances over virtual machines. The authors in [ 23 ] used YCSB to evaluate the Cassandra database on a Raspberry Pi 2 cluster.…”