Technologies Paper

ManualsBrandsIntel ManualsOtherIntel Xeon Processor 7020

Executive Summary

It’s not very often that a disruptive technology changes the way enterprises operate.

In the same way that the microprocessor, the personal computer, and virtualization

changed the computing landscape, Apache Hadoop* and the MapReduce framework

have forever changed the way that enterprises capture, store, and analyze information.

But while yesterday’s hardware technology helped lay the foundation for big data

analysis, today’s technologies let enterprises analyze more data faster than ever before.

Intel is leading the big data infrastructure charge with advancements in microprocessors,

storage, and networking. These advancements can help increase the scalability and

performance of large Apache Hadoop clusters. In internal tests that compared servers

equipped with the previous generation Intel® Xeon® processor E5 family and servers

equipped with the Intel® Xeon® processor E7 v2 family, the servers equipped with the

Intel Xeon processor E7 v2 family demonstrated performance gains of up to 3.5 times

across a spectrum of I/O- and CPU-intensive workloads.

This paper highlights technologies available from Intel that enterprises can use to scale

up Apache Hadoop clusters to handle the increasing volume, variety, and velocity of

data. Enterprises can reduce the complexity and total cost of ownership (TCO) of their

clusters by using fewer, more powerful servers, which can reduce operational costs up

to 37 percent overall over a four-year period.

Apache Hadoop* Overview

Apache Hadoop is a distributed data storage and data processing platform that

enterprises can use for storing and processing large amounts of semi-structured or

unstructured data. Built on Java*, Apache Hadoop has open-source roots and enjoys the

support of a large, active user and developer community. Apache Hadoop also benets

from the collaborative work of Java Virtual Machine (JVM) vendors and Intel engineers to

increase Java performance on the latest Intel platforms. These traits help make Apache

Hadoop a cost-effective, high-performance platform for enterprises to gather and

analyze data from such varied sources as point-of-sale systems, credit card transactions,

server log les, machine logs, and scientic sensors. Intel has worked with Java

Virtual Machine vendors for more than 10 years to optimize Java performance on Intel

hardware, as each new generation of Intel microarchitecture provides new features that

can increase software performance. All of this capability enables advanced analytics for

a range of tasks, from detecting credit card fraud to decoding the human genome.

An Apache Hadoop cluster can scale from a few servers to thousands. This exibility

makes Apache Hadoop an ideal platform across the data analysis spectrum, from

Accelerate Big Data Analysis

with Intel

Technologies

Tim Allen

Big Data Domain Expert

Intel Software & Services Group

Eric Kaczmarek

Big Data Performance Architect

Intel Software & Services Group

Frank Jensen

Performance Marketing Engineer

Intel Data Center Group Marketing

White Paper

Intel® Xeon® processor E7 v2

Big Data Analysis

Summary of content (7 pages)

PAGE 1
White Paper Intel® Xeon® processor E7 v2 Big Data Analysis Accelerate Big Data Analysis with Intel® Technologies Executive Summary It’s not very often that a disruptive technology changes the way enterprises operate. In the same way that the microprocessor, the personal computer, and virtualization changed the computing landscape, Apache Hadoop* and the MapReduce framework have forever changed the way that enterprises capture, store, and analyze information.
PAGE 2
Accelerate Big Data Analysis with Intel® Technologies Table of Contents Executive Summary. . . . . . . . . . . . . . . 1 Apache Hadoop* Overview. . . . . . . . 1 Performance Bottlenecks. . . . . . . . . . 2 Increase Apache Hadoop Cluster Performance with Intel® Technologies. . . . . . . . . . . . . .
PAGE 3
Accelerate Big Data Analysis with Intel® Technologies performance requirements—such as analyzing high-velocity data—while relieving performance bottlenecks within the cluster.
PAGE 4
Accelerate Big Data Analysis with Intel® Technologies Data Loading and Replication Performance • Apache* Hive Join, a workload that is both CPU- and I/O-intensive. This workload provides performance benchmarks for more structured datasets. • Page Rank, a MapReduce workload that uses a well-known search engine algorithm that ranks pages.
PAGE 5
Accelerate Big Data Analysis with Intel® Technologies With memory configurations up to 12 terabytes of RAM, enterprises can increase the amount of in-memory buffer used by MapReduce tasks, which can significantly reduce data movement between memory and disk-based storage. In servers with smaller memory configurations, MapReduce tasks can quickly fill the buffer, which means that the contents of the buffer are automatically moved—or “spilled”—out of memory and onto disk.
PAGE 6
Accelerate Big Data Analysis with Intel® Technologies TABLE 1: FOUR-YEAR OPERATIONAL COSTS COMPARISON BETWEEN A 1000-NODE INTEL XEON PROCESSOR E5-2680 CLUSTER AND A 334-NODE INTEL XEON PROCESSOR E7-4890 V2 CLUSTER. ANNUAL COSTS PER SERVER2 MAINTENANCE POWER AND COOLING YEARS 1 THROUGH 4 2-socket Intel® Xeon® processor E5-2680 (2.7 GHz, 8 cores) $500 $496 $155 $15 $1166 4-socket Intel® Xeon® processor E7-4890 v2 (2.
PAGE 7
Accelerate Big Data Analysis with Intel® Technologies 1 Based on Intel Corporation internal testing. The Intel® Xeon® E5-2680 Apache Hadoop cluster consisted of a master node and five worker nodes. All nodes were configured with dual Intel Xeon E5-2680 processors running at 2.7 GHz, 128 GB of RAM (8x16 GB DDR3 1600 Mhz), 3 Intel SSD DC S3700 Series SSDs (400GB, 2.5in SATA 6Gb/s, 25nm, MLC, Seq. Read: 500 MB/s, Seq. Write 460 MB/s), 3 Intel SSD 710 Series SSDs (200GB, 2.5in SATA 3Gb/s, 25nm, MLC, Seq.