This document discusses big data and its implications for data center architecture. It provides examples of big data use cases in telecommunications, including analyzing calling patterns and subscriber usage. It also discusses big data analytics for applications like genome sequencing, traffic modeling, and spam filtering on social media feeds. The document outlines necessary characteristics for data platforms to support big data workloads, such as scalable compute, storage, networking and high memory capacity.
Gen AI in Business - Global Trends Report 2024.pdf
Big Data Implications on Platform Architecture
1. Big Data and Implications on
Platform Architecture
Fayé A Briggs, PhD
Intel Fellow and Chief Server Platform Architect, Intel
BIGS002
2. Agenda
• What is Big Data?
• Big Data use cases
• What does Big Data mean for the data center?
• Call to action
The PDF for this Session presentation is available from our
Technical Session Catalog at the end of the day at:
intel.com/go/idfsessionsBJ
URL is on top of Session Agenda Pages in Pocket Guide
2
3. Agenda
• What is Big Data?
• Big Data use cases
• What does Big Data mean for the data center?
• Call to action
3
4. What is Big Data?
Unstructured size is beyond volume, variety, value and velocity
datasets whose the ability
of typical database software tools to capture, store, manage and analyze†
Unstructured
Data Analyze
Big Sensed Data
Manage
Volume
Big Corporate Data
Big Web Data
Store
Structured
Data
Corporate Data
Time
Capture
4 †”Big data: The next frontier for innovation, competition, and productivity”, McKinsey Global Institute
5. The Four Pillars and Associated
Challenges of Big Data
Massive scale and growth of unstructured data
80%~90% of total data
Volume Growing 10x~50x faster than structured (relational) data
10x~100x of traditional data warehousing
Heterogeneity and variable nature of Big Data
Many different forms (text, document, image, video, ...)
Variety No schema or weak schema
Inconsistent syntax and semantics
Predictive analytics for future trends and patterns
Value Deep, complex analysis (machine learning, statistic modeling,
graph algorithms, …), versus
Traditional business intelligence (querying, reporting, …)
Real-time rather than batch-style analysis
Velocity Data streamed in, tortured, and discarded
Making impact on the spot rather than
after-the-fact
5
6. Big Data Has Gone Mainstream
“ . . . . every large company is working on Big Data projects “at a
furious pace.” - Gartner analyst Merv Adrian
“. . . the use of Big Data will be effective for every segment of the
economy.” - Michael Chui, McKinsey & Co. analyst
Behold the Big Data sign Greeting commuters on Highway 101 in Silicon
gracing Times Square Valley, a giant Walmart* Big Data sign
6
7. Big Data Usages Examples
(Telecom/Financial/Search)
• Telecom
– Calling patterns, signal processing, forecasting
Analyze switches/routers data for quality of call, frequency of
calls, region loads, etc.
– Act before problems happen. Act before customer calls arrive.
• Financial
– Trading behaviour
Analyze real-time data to understand market behavior, role of
individual institution/investor
– Detect fraud, detect impact of an event, detect the players
• Search Engines
– Process the data collected by Web bot in multiple dimensions
– Enhance relevance of search
Big Data impacts e-connected businesses through capture,
processing and storage of huge amount of data efficiently
7
8. Big Data Usages Examples
(E-Biz, Media)
• Click Stream Analysis
– Analysis of online users behavior
– Develop comprehensive insight (Business Intelligence) to run
effective strategies in real time
• Graph analysis
– Term for discovering the online influencers in various walks of life
– Enables a business to understand key players and devise effective
strategies
• Lifecycle Marketing
– Strategies to move away from spam/mass mail
– Enables a business to spend money on high probable customers only
• Revenue Attribution
– Term for analyzing the data to accurately attribute revenue back to
various marketing investments
– Business can identify effectiveness of campaign to control expenses
Big Data phenomenon allows businesses to know,
predict and influence customer behaviors!!!
8
9. Big Data in Health Care
Cancer Care: American Society of Clinical
Oncologists “learning health system”, CancerLinQ
• Benefits: Collects and analyzes de-anonymized cancer
care data from millions of patient visits
Genome Sequencing: Improvements in DNA
sequencing driving down costs of processing
complete set of genomes
• Benefits: Saving lives through better identification and
treatment of diseases
GenBank* is the NIH genetic sequence
database, an annotated collection of all
publicly available DNA sequences
Sequencing costs approaches “$1000.” Analytics,
compute, networking, & storage to improve
affordability still challenging.
9
10. Big Data Analytics Processing
• “Batch”: Sophisticated data processing: enable “better”
decisions
– Analyze, transform, scan, etc. large amount of data
– E.g., ETL, graph construction, anomaly detection, trend analysis
• “Real-time”: Queries on historical data: enable “faster”
decisions
– Data at rest but needs to be served in real-time
– E.g., Facebook* uses HBase* “messaging” App serving real-time
data to its users
• “Streaming”: Queries on live data: enable “instantaneous”
decisions on real-time streaming data
– Large volume of data being ingested and analyzed in real-time
– E.g., detect and block worms in real-time (a worm may infect
1mil hosts in 1.3sec)
10 Adapted from Ion Stoica, UC Berkeley
11. Big Data Converted to Knowledge in an Iterative Cycle†
Knowledge Deliver
Visualization
Presentation
Tableau*, R, Progress Software*, Pentaho*, IBM*, others
Transact
Batch Analytics Real-time Analytics Streaming Analytics
Call Data Records Intelligent Transport Twitter* Spam Analytics
Gene Seq & Analysis Home Energy Video meta-data
GraphLab*-MLearning, Financial Analytics Traffic Modeling
ETL, Sensor Network Database Virus Intrusion Detection
Search, Time Series Stock Ticks Stock trading
Index Creation, Customer Behavior Ads Video Surveillance Analytics
Click Stream Analysis, Retail – Video, PoS data
BI Analytics Compute
Batch Processing Real-Time Processing Stream Processing
MR-Hadoop*, HBase*, SAP* HANA, Spark, Spark Streaming, Storm,
GraphLabs,Giraph Shark, Cassandra*, S4–Simple Scalable Stream.
mongoDB, Drill, Impala Sys
Data
Management
External Unstructured/Semi-structured on Dist. Parallel File System
(HDFS*, Lustre*, CloudStore*, GPFS, GlusterFS*, etc.)
Archive
Filtering, Cleansing, ETL, Scribe, etc.
Big Data Ingest
†Based on Frog’s & IDC’s Layered & Iterative Approach
11
12. Agenda
• What is Big Data?
• Big Data use cases
• What does Big Data mean for the data center?
• Call to action
12
13. Telco Usage - China Mobile Group Guangdong
Hadoop* Big Data storage and analytics
Analytics
13
14. Telco Usage - China Mobile Group Guangdong
Hadoop* Big Data storage and analytics
Usage Model: Deliver real-time access to Call Data Records (CDR)
for billing self service
• Solution: Hadoop* + Intel® Xeon® processors over RDBMS to remove
data access bottlenecks, increase storage, and scale system
• Benefits: Lower TCO, up to 30x performance increase, stable operation,
analytics on subscriber usage for targeted promotions
• Characteristics:
– 30TB billing data/month; real-time retrieval of 30 days CDRs
– 300k records/sec, 800k insert speed/sec; 15 analytics queries , 133 server nodes+
Platform and Cluster Architectural Attributes:
• Compute:
– Scale-out Intel Xeon processor E5 based platform for fast analytic query processing
– Memory: 2-4GB/Core for Data Management Hadoop JVM
– PCI Express* Gen3
• Storage: Scale-out Storage (HDD) for capacity; SSD Cache
• I/O: High bandwidth and IOPS to Storage (HDD + SSD cache) for fast record inserts
and reads
• Network: High network bandwidth for Hadoop Shuffle data; 10GbE TORS; 10-40GbE
14 Inter-Rack Switch
15. Real-Time Analytics: DuPont*– Crop Genetics
HBase* Big Data analytics with “BLAST” to compare protein in
genomic data
Usage Model: Comparative genomic research. Run “BLAST” to compare
protein with every other protein in the genomic code.
• Solution: The current RDBMS didn’t scale to planned data growth. HBase*/HDFS*
proved a reduction in processing time from over 30 days to less than 7 hours.
• Characteristics:
– Genetic data for 4 million organisms – 1TB in size; scale to 14 million
– 12 Trillion HBase rows; single record search takes 1.2 seconds
Platform and Cluster Architectural Attributes:
• Compute:
– Scaleout Intel® Xeon® processor E5 base platform for real-time data serving
– Memory: Higher Memory Capacity 4GB/Core for HBase Memstores
– PCI Express* Gen3
• Storage: Scaleout Storage (HDD) for capacity; SSD Cache
• I/O: High bandwidth and IOPS to Storage (HDD + SSD cache) for fast data
access to HBase tables
• Network: High network bandwidth for HBase region servers; 10GbE TORS; 10-
40GbE Inter-Rack Switch
15
16. Telco - China Unicom
Hadoop* & HBase* for Behavioral Analysis
Subscriber Usage &
Billing
ETL
Storage, Analytics
• Log Analysis
• Daily Reports
New Customer Segmentation & Insights
16
17. Telco - China Unicom
Hadoop* & HBase* for Behavioral Analysis
Usage Model: Analyze subscriber Web usage and billing to derive new information
products
• Solution: Scale out storage based on Hadoop* & HBase* with network optimization based on
Web traffic, log analysis for daily reporting
• Benefits: New customer segmentation
• Characteristics:
– 188 nodes, 14TB/server
– 2.5PB raw disk capacity
– High speed data loading
– Real-time query (latency <1s)
Daily statistics & reports (sum, count, join, etc.)
Platform and Cluster Architectural Attributes:
• Compute:
– Scaleout Intel® Xeon® processor E5 based platform for real-time data serving in < 1 sec
– Memory: Higher Memory Capacity 4GB/Core for HBase Memstores
– PCI Express* Gen3
• Storage: Scaleout Storage (HDD) for capacity; SSD Cache
• I/O: High bandwidth and IOPS to Storage (HDD + SSD cache) for fast data
access to HBase tables
• Network: High network bandwidth for HBase region servers; 10GbE TORS; 10-
40GbE Inter-Rack Switch
17
19. In-Memory GraphLab* Analytics: PageRank
Big Data analytics with GraphLab
Usage Model: Deliver Page Ranking for search
• Solution: Hadoop* + Intel® Xeon® processors: Large number of ML, Genomics,
Web, etc. applications can be efficiently run in Graph Parallel solution
• Benefits: Significantly faster solutions to Graph(Vertices, Edges) domain
problems
• Characteristics:
– XML docs, News Feeds, Web Pages
– Data collected from Web Pages for Page Ranking
Platform and Cluster Architectural Attributes:
• Compute:
– Scaleout Intel Xeon processor E5 based platform for fast analytic in-memory processing
– Memory: 2-4GB/Core for Data Management Hadoop JVM; For graph data
– PCI Express* Gen3
• Storage: Scaleout Storage (HDD) for capacity; SSD Cache
• I/O: High bandwidth and IOPS to Storage (HDD + SSD cache) for fast record inserts and
reads
• Network: High network bandwidth for Hadoop Shuffle data during graph construction;
10GbE TORS; 10-40GbE Inter-Rack Switch
19
20. Pipelined In-Memory Analytics: Twitter* Feed Spam Analytics
Spark Streaming Big Data Analytics
• Run a streaming computation as a series of
extremely small, deterministic batch jobs
• Batch sizes as low as ½ second, latency ~ 1 second
*
20
21. Pipelined In-Memory Analytics: Twitter* Feed Spam Analytics
Spark Streaming Big Data Analytics
Usage Model: Process and filter out Twitter* feed spam as tweets are
ingested
• Solution: Spark Streaming from Berkeley
• Characteristics:
– Data ingested from Twitter feeds
Platform and Cluster Architectural Attributes:
• Compute:
– Scaleout Intel® Xeon® processor E5 based platform for fast analytic query processing
– Memory: Very large Memory Capacity close to the CPU; High memory bandwidth
– PCI Express* Gen3
• Storage: Scaleout Storage (HDD) for capacity; SSD Cache
• I/O: High bandwidth and IOPS to Storage (HDD + SSD cache) for fast record
inserts and reads
• Network: High network bandwidth; 10GbE TORS; 10-40GbE Inter-Rack
Switch
21
22. Smart City Sensor Model
Embedded
Smart Smart Grid
building sensors Cloud
Industrial
sensors automation
Pollution sensors
sensors Dedicated
Meteorological Smart
sensors meters
HPC
INTELLIGENT INTELLIGENT
CITY FACTORY
Transactional
INTELLIGENT INTELLIGENT
HOSPITAL HIGHWAY
Social
Sensors on
Inductive Traffic
Portable medical Medical smartphones
Sensors on sensors cameras
imaging services sensors on
vehicles Location
ambulances
22
23. The Fusion of Internet of Things and Big Data
Planogram monitoring Real-time transaction
(Real-time stock level) Dynamic pricing
Real-time, personalized ad
Auto promotion/coupon
Social network connection
Interactive display
(Behavioral marketing)
RFID
RFID
Store heat map (hot merchandise, browsing
Surveillance camera (store statistics) history, conversion rate)
23
25. Smart Traffic Intelligent Transport System
HBase* Application for Predictive Analytics
Usage Model: Analyze city traffic to derive statistics for crime prevention, info sharing, and
predictive traffic analysis
• Solution: Embed HBase* client in camera for real-time inserts of structured/unstructured
data
• Benefits: Automated queries for traffic violation, data mining of fake licenses <1 minute for
all data captured for a week, predictive traffic forecasting
• Characteristics:
– 30000 + camera data collection points
– Petabytes of traffic data & terabytes of images
– 2 billion HBase records
Platform and Cluster Architectural Attributes:
• Compute:
– Scaleout Intel® Xeon® processor E5 based platform for real-time data serving
– Memory: Higher Memory Capacity 4GB/Core for HBase Memstores
– PCI Express* Gen3
• Storage: Scaleout Storage (HDD) for capacity; SSD Cache
• I/O: High bandwidth and IOPS to Storage (HDD + SSD cache) for fast data
access to HBase tables
• Network: High network bandwidth for HBase region servers; 10GbE TORS; 10-
40GbE Inter-Rack Switch
25
26. Video created
Video analyzed
Video Cold storage
E2E Analytics Use Case - Safe City
Video metadata
stored
Camera
Video Storage
(Edge or Centralized)
Private Cloud Public Cloud
2-3 TB Video
data/Camera/Month
Management
System
Police Car
300 GB
Edge Device metadata/Camera/Month
Edge Client Video
(Video capture)
Data Center/ Data Services
Indexer/Analyzer/Transcoder Cloud (VSaaS, VAaaS)
Smart
(Image extraction & Metadata Creation) (Private/Public)
Checkpoint District City State Country
By end of 2017 Edge & Backend VA’s Value By end of 2017
457 PB Metadata tagging and compression at the edge 76 PB
Raw Video per Metadata per
Enables 10s of new use models for traffic mgmt
Day Day
Traffic video People, cars,
Fastest time to information and public safety
streamed by HD geospatial
cameras information
Typical Scenario : Automated traffic violations
26
E.g., 70% Traffic violation detection by video in 2011
27. Smart City Big Data Architecture Framework
Intel® Many Visualization & Interpretation
Integrated Core
Architecture
Vertical Scale
Based on Intel®
Horizontal &
Streaming [Un]Structured Batch
microarchitecture-EX Analytics Data Analytics
Based on Intel
Data Acquisition
microarchitecture-EP
Microserver
Local Analytics
Based on Intel
Complex Event Processing
microarchitecture-EN Analytics Processing
Horizontal
Preprocessing/
Storage
Cleansing/Filtering/
Intel® Core™
Scale
Aggregation
Data Acquisition Video Analytics
Sensors Cameras
Core System-
on-a-Chip
27
28. Agenda
• What is Big Data?
• Big Data use cases
• What does Big Data mean for the data center?
• Call to action
28
29. Choice of Compute Platforms Optimized for Big Data
Intel® Xeon® processor E5 Family Intel Xeon processor E7 Family
RAM
QPI 1 Xeon E7-4800
QPI 2 CORE 1 CORE 2
QPI 3 CORE 3 CORE 4
CORE 5 CORE 6
QPI 4
Up to 4 channels CORE 7 CORE 8 Up to 8 channels
Integrated DDR3 1600 MHz DDR3 1066 MHz
PCI Express* memory CORE 9 CORE 10 memory
3.0
4 QPI 1.0
Up to 40
Up to 8 cores Lanes for
robust
CACHE Up to 10 cores
Up to 20 MB Up to 30 MB
lanes cache scalability
per socket cache
• Preferred solution for Hadoop* and scale- • Preferred solution for in-memory analytic
out analytic/DW engines engines and enterprise databases
• Up to 80%** performance boost compared • Highest cache and thread performance for
to prior generation large-dataset processing
• Intel® Integrated I/O with PCI Express* • Up to 2TB memory footprint (4-socket
3.0 provides more bandwidth for large platform) for in-memory apps
data sets • Highest reliability and 8-socket+ scalability
• Latest DDR3 memory technology/capacity
for reduced memory latency
Right Analytic Platforms begins with Intel Xeon processors
29 QPI = Intel® QuickPath Interconnect. **See backup slides for 80% claim
30. Platform and Software Optimizations for Hadoop*
Integrated Up to four channels
PCI Express* DDR3 1600 MHz
3.0 memory
Up to 40 Up to eight
lanes cores
per socket
Up to 20 MB
cache •
• Up to 80%** performance boost vs. prior generation
– Intel® Advanced Vector Extensions - reduce compute time
– Intel® Turbo Boost Technology - increased performance
• Intel® Distribution for Apache Hadoop* software
– Built on open source releases
– Custom tuning for data types and scaling approaches
** See backup slides for 80% claim
1 Performance comparison using best submitted/published 2-socket server results on the SPECfp*_rate_base2006 benchmark as of 6 March 2012.
2 Source: Intel internal measurements of average time for an I/O device read to local system memory under idle conditions comparing Intel® Xeon® processor E5-2600 product family (230 ns) vs..
QPI = Intel® QuickPath Interconnect
Intel® Xeon® processor 5500 series (340 ns). See notes in backup for configuration details
Intel® Xeon® processor E5
30 * Other names and brands may be claimed as the property of others
32. Microserver: High Density, Low
Power System Innovations
• Addressing the low power, high
density packaging
• Based on Intel® Atom™ processors
– Next generation Intel Atom
processor codename Avoton
– Workloads
Web tier, SaaS, IaaS, PaaS and
light data analytics
– For scale-out apps
32
33. Transforming Storage
Data explosion … Driving storage opportunity
690% Distributed 30%
CAGR
Growth in storage capacity 2010- Storage
2015+
Traditional 16%
Volume Unstructured Data Storage CAGR
Big Sensed Data
Intel® Xeon® processors provide
Big Corp Data
storage intelligence
Big Web Data • Deduplication
Structured Data • Thin
provisioning
Corporate Data • Erasure code
• MapReduce
Time • Encryption
690 percent growth in storage capacity based off Intel analysis and IDC data,
between 2010 (26,066 petabytes) to 2015 (179,327) which is ~690%
Source: Intel
33
34. Intelligent Distributed Storage
Optimizations
Intelligent pattern matching reduces large
blocks of repeated data
BEFORE AFTER
TRADITIONAL
Real-Time Data Analytics
THIN PROVISIONING
ALLOCATION
ALLOCATED -FREE
On-demand utilization of available storage –
APPLI 2 SYSTEM-WIDE
APPLI 1
ALLOCATED - USED
ALLOCATED -FREE
CAPACITY RESERVED
APPLI 2
virtual and real capacity
ALLOCATED -USED APPLI 1
Analysis of real-time storage determines
extent and nature of compression
Strategic positioning of faster storage
devices, improves storage performance
34
35. Agenda
• What is Big Data?
• Big Data use cases
• What does Big Data mean for the data center?
• Call to action
35
36. Call to Action
• Big Data represents a huge industry opportunity for
innovation – get involved!
– New solutions for analytics
– Hardware infrastructure innovation – across the platform
• Customers from across enterprise, cloud,
government, HPC and telecom are looking to improve
decision making with big data
– If you are a developer of solutions: Understand the market
opportunities
– If you are a manager of solutions: Understand where big
data can help your organization
• Intel is deeply engaged in big data
– Work with us on delivery of big data solutions
36
38. Legal Disclaimer
• Intel® Turbo Boost Technology requires a system with Intel Turbo Boost Technology. Intel Turbo Boost Technology and
Intel Turbo Boost Technology 2.0 are only available on select Intel® processors. Consult your PC
manufacturer. Performance varies depending on hardware, software, and system configuration. For more information,
visit http://www.intel.com/go/turbo.
38
39. Risk Factors
The above statements and any others in this document that refer to plans and expectations for the first quarter, the year and the
future are forward-looking statements that involve a number of risks and uncertainties. Words such as “anticipates,” “expects,”
“intends,” “plans,” “believes,” “seeks,” “estimates,” “may,” “will,” “should” and their variations identify forward-looking
statements. Statements that refer to or are based on projections, uncertain events or assumptions also identify forward-looking
statements. Many factors could affect Intel’s actual results, and variances from Intel’s current expectations regarding such factors
could cause actual results to differ materially from those expressed in these forward-looking statements. Intel presently considers the
following to be the important factors that could cause actual results to differ materially from the company’s expectations. Demand
could be different from Intel's expectations due to factors including changes in business and economic conditions; customer acceptance
of Intel’s and competitors’ products; supply constraints and other disruptions affecting customers; changes in customer order patterns
including order cancellations; and changes in the level of inventory at customers. Uncertainty in global economic and financial
conditions poses a risk that consumers and businesses may defer purchases in response to negative financial events, which could
negatively affect product demand and other related matters. Intel operates in intensely competitive industries that are characterized by
a high percentage of costs that are fixed or difficult to reduce in the short term and product demand that is highly variable and difficult
to forecast. Revenue and the gross margin percentage are affected by the timing of Intel product introductions and the demand for and
market acceptance of Intel's products; actions taken by Intel's competitors, including product offerings and introductions, marketing
programs and pricing pressures and Intel’s response to such actions; and Intel’s ability to respond quickly to technological
developments and to incorporate new features into its products. The gross margin percentage could vary significantly from
expectations based on capacity utilization; variations in inventory valuation, including variations related to the timing of qualifying
products for sale; changes in revenue levels; segment product mix; the timing and execution of the manufacturing ramp and
associated costs; start-up costs; excess or obsolete inventory; changes in unit costs; defects or disruptions in the supply of materials
or resources; product manufacturing quality/yields; and impairments of long-lived assets, including manufacturing, assembly/test and
intangible assets. Intel's results could be affected by adverse economic, social, political and physical/infrastructure conditions in
countries where Intel, its customers or its suppliers operate, including military conflict and other security risks, natural disasters,
infrastructure disruptions, health concerns and fluctuations in currency exchange rates. Expenses, particularly certain marketing and
compensation expenses, as well as restructuring and asset impairment charges, vary depending on the level of demand for Intel's
products and the level of revenue and profits. Intel’s results could be affected by the timing of closing of acquisitions and divestitures.
Intel’s current chief executive officer plans to retire in May 2013 and the Board of Directors is working to choose a successor. The
succession and transition process may have a direct and/or indirect effect on the business and operations of the company. In
connection with the appointment of the new CEO, the company will seek to retain our executive management team (some of whom are
being considered for the CEO position), and keep employees focused on achieving the company’s strategic goals and objectives. Intel's
results could be affected by adverse effects associated with product defects and errata (deviations from published specifications), and
by litigation or regulatory matters involving intellectual property, stockholder, consumer, antitrust, disclosure and other issues, such as
the litigation and regulatory matters described in Intel's SEC reports. An unfavorable ruling could include monetary damages or an
injunction prohibiting Intel from manufacturing or selling one or more products, precluding particular business practices, impacting
Intel’s ability to design its products, or requiring other remedies such as compulsory licensing of intellectual property. A detailed
discussion of these and other factors that could affect Intel’s results is included in Intel’s SEC filings, including the company’s most
recent Form 10-Q, report on Form 10-K and earnings release.
Rev. 1/17/13
39
41. Disclaimer for “Up to 80% performance
boost compared to prior generation”
• Performance comparison using best submitted/published
2-socket server results on the SPECfp*_rate_base2006
benchmark as of 6 March 2012. Baseline score of 271
published by Itautec on the Servidor Itautec MX203* and
Servidor Itautec MX223* platforms based on the prior
generation Intel® Xeon® processor X5690. New score of
492 submitted for publication by Dell on the PowerEdge
T620 platform and Fujitsu on the PRIMERGY RX300 S7*
platform based on the Intel® Xeon® processor E5-2690.
For additional details, please visit www.spec.org. Intel
does not control or audit the design or implementation of
third party benchmark data or Web sites referenced in
this document. Intel encourages all of its customers to
visit the referenced Web sites or others where similar
performance benchmark data are reported and confirm
whether the referenced benchmark data are accurate and
reflect performance of systems available for purchase.
41