• Like

Capturing Big Data for Rapid Analytics

  • 250 views
Uploaded on

Timely analysis of information derived from big data volumes depends largely on how fast the data can be accessed. Dell, Samsung and Microsoft put access speed to the test using advanced servers with …

Timely analysis of information derived from big data volumes depends largely on how fast the data can be accessed. Dell, Samsung and Microsoft put access speed to the test using advanced servers with energy-efficient memory and solid-state drives.

More in: Business
  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Be the first to comment
    Be the first to like this
No Downloads

Views

Total Views
250
On Slideshare
0
From Embeds
0
Number of Embeds
0

Actions

Shares
Downloads
2
Comments
0
Likes
0

Embeds 0

No embeds

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
    No notes for slide

Transcript

  • 1. 22 2013 Issue 02 | dell.com/powersolutionsBusiness intelligenceSpecialsectionInnovative devices, sensors, streamingvideo and many other sources areaccelerating the volume, variety andvelocity of data that is ripe for big dataanalytics. Characterized by enormous datasets beyond the capabilities of conventionalsoftware tools to capture, manage andprocess within acceptable time periods,big data offers unprecedented insight intostrategic planning and growth opportunities.As a result, IT leaders across many industriesare responding to the competitive urgencyfor launching big data initiatives that tapinto massive repositories of structured,semi-structured and unstructured data.At the outset, many IT organizationsare discovering that they do not havethe computing technology or power toadequately collect, sort and categorize datamined from stores containing terabytes,petabytes and even exabytes of data. Tocapitalize on these rich information sources,organizations require scalable technologiesthat enable them to store, manage and accessdata designed to improve business intelligencewithout increasing operating expenses.Accelerating data captureIn today’s business environment wheredecisions are based on dispersed, dynamicinformation sources, access to big data foranalytics must keep pace. An in-memoryapproach expedites data capture by loadingan entire database into main memory onthe database server. Access to data storedin a database server’s RAM is substantiallyfaster than access to data stored on harddisk drives (HDDs) — typically measured innanoseconds, versus milliseconds.Another highly efficient alternative isflash-based storage or solid-state drives (SSDs)that are designed to significantly increase theperformance of servers, such as those usedfor big data tasks. And compared to HDDs,SSDs help reduce power consumption.Moreover, SSDs are significantly faster, withaccess time in microseconds, than traditionaldisk-based arrays — even high-performingarrays. They are designed to provide anoptimal solution for organizations that wantrapid access to frequently used data takendirectly from memory — or a warm cache —and access to data used less frequently fromdisk drives — or a cold cache.Dell™ PowerEdge™ servers — equippedwith fast, power-efficient Samsung® GreenDouble Data Rate 3 (DDR3) dynamic RAM(DRAM) and Samsung Green SSDs — canrun database software capable of handlinglarge amounts of warm and cold cache data.These servers, in particular, offer scalableplatforms for database-intensive applicationsand help deliver the bandwidth needed formoving data quickly across the enterprise.Comparing warm and cold cache accessTo test data access and analysis capabilities,Samsung and Microsoft performed proof-of-concept (POC) testing in June 2012 atthe Microsoft Technology Center (MTC) inTimely analysis of information derived from big datavolumes depends largely on how fast the data can beaccessed. Dell, Samsung and Microsoft put access speedto the test using advanced servers with energy-efficientmemory and solid-state drives.By Peyman Blumstengel, Marc Crespell, Sylvie Kadivar,Yogev Shimony and Jessica TitusCapturing big datafor rapid analyticsReprinted from Dell Power Solutions, 2013 Issue 2. Copyright © 2013 Dell Inc. All rights reserved.
  • 2. dell.com/powersolutions | 2013 Issue 02 23Paris, France. In addition to powerconsumption efficiency, the testingmeasured the duration of queries in fourscenarios that represent key data access andanalysis tasks using the TCP Benchmark™H(TCP-H) decision support benchmark.The test configuration consisted of twoDell PowerEdge R910 servers poweredby Intel® Xeon® E7-4870 processors withSamsung Green DDR3 DRAM. One serverwas set up with 1 TB Samsung Green20 nm–class DDR3-1066 memory modulesin a high-performance, energy-efficientconfiguration; the other server wasequipped with mainstream, 50 nm–classDDR3-1066 memory modules in a standardconfiguration. Each server was set up to dualboot Microsoft® Windows Server® 2008 R2Service Pack 1 (SP1) and Windows Server 2012Release Candidate operating systems, andeach ran the Microsoft® SQL Server® 2012database engine. The high-performance,energy-efficient configuration includedSamsung Green SSDs and Samsung PM830Serial ATA (SATA) SSDs for a total capacity of4 TB. The standard configuration wasequipped with 15,000 rpm Serial AttachedSCSI (SAS) HDDs for a total capacity of 3.2 TB.The test results demonstrated that the high-performance, energy-efficient configuration was15 times faster than the standard configurationfor data access in a cold cache environment.This configuration saved 94 percent of systempower consumption (see figure).The high-performance, energy-efficientsystem demonstrated 2 percent faster dataaccess than the standard configuration in awarm cache environment in which almostall the data was located in memory. Thisconfiguration saved 30 percent of systempower consumption (see figure).Mining big data quickly and efficientlyIn this test study, a Dell PowerEdge R910server in a high-performance, energy-efficientconfiguration with Samsung Green DDR3and Samsung Green SSDs running Microsoftsoftware demonstrated a data access speedwell suited for big data analytics. Thesetest results indicate that the POC testconfiguration is designed to deliver rapidaccess, management and processing ofdata that can be mined for enterprise-scalebig data analytics within a cost-efficientpower envelope.1,4001,2001,0008006004002000WattsSource: “Big Data Implementation: Role of Memory and SSD in Microsoft SQL Server Environment,” Microsoft Technology Center (MTC) and SamsungElectronics, September 2012.0 30 60 90 120 150 180Minutes50 nm–class DDR3 and HDD on Windows Server 2012 Release Candidate20 nm–class DDR3 and SSD on Windows Server 2012 Release CandidateHigh-performance and energy-efficient server: 11 minutes, 8 seconds at 126 WhCold-cache data access performance and power consumption comparison between high-performance,energy-efficient and standard configurations1,4001,2001,0008006004002000WattsSource: “Big Data Implementation: Role of Memory and SSD in Microsoft SQL Server Environment,” Microsoft Technology Center (MTC) and SamsungElectronics, September 2012.0 30 60 90 120 150 180Minutes50 nm–class DDR3 and HDD on Windows Server 2012 Release Candidate20 nm–class DDR3 and SSD on Windows Server 2012 Release CandidateHigh-performance and energy-efficient server: 1 minute, 45 seconds at 23 WhWarm-cache data access performance and power consumption comparison between high-performance,energy-efficient and standard configurationsAuthorsPeyman Blumstengel is a green evangelistat Samsung Semiconductor Europe.Marc Crespell is a global alliances managerat Dell.Sylvie Kadivar is a senior director of strategicmarketing at Samsung Semiconductor Inc.Yogev Shimony is a business developmentmanager in Global Partner Development at Dell.Jessica Titus is a senior manager of strategicmarketing at Samsung Semiconductor Inc.Learn moreDell PowerEdge servers:Dell.com/poweredgeSamsung Green Memory:www.samsung.com/greenmemorySamsung IT ecosystems:qrs.ly/e83bu4jReprinted from Dell Power Solutions, 2013 Issue 2. Copyright © 2013 Dell Inc. All rights reserved.