A performance study of big data on small nodes

Abstract

The continuous increase in volume, variety and velocity of Big Data exposes datacenter resource scaling to an energy utilization problem. Traditionally, datacenters employ x86-64 (big) server nodes with power usage of tens to hundreds of Watts. But lately, low-power (small) systems originally developed for mobile devices have seen significant improvements in performance. These improvements could lead to the adoption of such small systems in servers, as announced by major industry players. In this context, we systematically conduct a performance study of Big Data execution on small nodes in comparison with traditional big nodes, and present insights that would be useful for future development. We run Hadoop MapReduce, MySQL and in-memory Shark workloads on clusters of ARM big.LITTLE boards and Intel Xeon server systems. We evaluate execution time, energy usage and total cost of running the workloads on selfhosted ARM and Xeon nodes. Our study shows that there is no one size fits all rule for judging the efficiency of executing Big Data workloads on small and big nodes. But small memory size, low memory and I/O bandwidths, and software immaturity concur in canceling the lower-power advantage of ARM servers. We show that I/O-intensive MapReduce workloads are more energy-efficient to run on Xeon nodes. In contrast, database query processing is always more energy-efficient on ARM servers, at the cost of slightly lower throughput. With minor software modifications, CPU-intensive MapReduce workloads are almost four times cheaper to execute on ARM servers.

A Study on the Energetic Viability of Single Board Computers for Cloud Computing Scenarios

“…As stated in previous works [14], ARM-based architectures are not univocally superior to traditional datacenter infrastructures neither in raw data processing nor in standard energy usage. In a first instance this seems to oppose the optimistic findings of Svandeldt-Winter [15], but we must consider the excellent improvements to x86 energy state management in the last few years, which have Figure 7.…”

Energy Proportionality in Heterogeneous Data Center Supporting Applications with Variable Load

“…Their performance has improved over time, however, they are not yet as powerful as traditional data center servers. Existing work has studied the use of low power processors for handling data center workloads like web servers [14] and big data [15]. It has been shown that popular Raspberry Pi's are efficient for hosting static web servers while consuming less power than a standard server [14].…”

“…It has been shown that popular Raspberry Pi's are efficient for hosting static web servers while consuming less power than a standard server [14]. ARM processors also offer interesting consumptionperformance ratios for database query processing compared to an Intel Xeon processor [15]. On the other hand, these studies show performance limitations, concluding that these equipments cannot compete with standard servers for more demanding workloads such as dynamic web servers or I/O intensive big data applications.…”

An experimental evaluation of datacenter workloads on low-power embedded micro servers

“…The first category is to seek energy proportionality with non-energyproportional servers [53,32,31,27,35]. The second category is to build more energy-efficient architecture based on low-power CPU [38,41,49,33,43,46,29].…”

“…Some of these non-traditional server platforms are equipped with low-power processors, such as ARM-based CPU [38,41,49], Intel Atom CPU [33,43,46,29,25] or even embedded CPU [21,50]. In addition to low-power CPU, more energyefficiency can be achieved by exploiting low-power flash storage [25,21] or highly customizable FPGA [40].…”