testing big data: automated testing of hadoop with querysurge
TRANSCRIPT
built by
Bill HaydukCEO/President
RTTS
Testing Big Data: Automated ETL Testing of Hadoop
Jeff Bocarsly, Ph.D.Chief Architect
QuerySurge Division, RTTS
built by
QuerySurge ™
Automate your Data Warehouse & Big Data Testing and Reap the Benefits
built by
QuerySurge ™
Today’s Agenda
• About Big Data and Hadoop
• Data Warehouse refresher
• Hadoop and DWH Use Case
• How to test Big Data
• Demo of QuerySurge & Hadoop
AGENDA
Topic: Testing Big Data: Automated ETL Testing of Hadoop
Host: RTTS
Date: Thursday, January 30, 2014
Time: 1:00 pm, Eastern Standard
Time (New York, GMT-05:00)
Session number:630 771 732
built by
QuerySurge ™
About FACTS
Founded: 1996
Locations: New York (HQ), Atlanta, Philadelphia, Phoenix
Strategic Partners:IBM, Microsoft, HP, Oracle, Teradata, HortonWorks, Cloudera, Amazon
Software:
QuerySurge
RTTS is the leading provider of software & data quality for critical business systems
built by
Facebook handles 300 million photos a day and about 105 terabytes of data every 30 minutes.
- TechCrunch
The big data market will grow from $3.2 billion in 2010 to $32.4 billion in 2017.- Research Firm IDC
65% of…advanced analytics will have Hadoop embedded (in them) by 2015.-Gartner
built by
QuerySurge ™
ETL
Source Data ETL Process Data WarehouseBig Data
Business Intelligence (BI) software
CxOs are using Business Intelligence & Analytics to make critical business decisions – with the assumption that the underlying data is fine.
“The average organization loses $8.2 million annually through poor Data Quality.”
- Gartner
Data Architecture
The Executive Office and Big Data
potential problem areas
Big data – defined as too much volume, velocity and variety to work on normal database architectures.
SizeDefined as 5 petabytes or more 1 petabyte = 1,000 terabytes 1,000 terabytes = 1,000,000 gigabytes1,000,000 gigabytes = 1,000,000,000 megabytes
about Big Data
built bybuilt by
QuerySurge ™
Big Data Impact
Handles more than 1 million customer transactions every hour.• data imported into databases that contain > 2.5 petabytes of data • the equivalent of 167 times the information contained in all the books in the US Library of
Congress.
Facebook handles 40 billion photos from its user base.
Google processes 1 Terabyte per hour
Twitter processes 85 million tweets per day
eBay processes 80 Terabytes per day
others
built by
QuerySurge ™
Requires exceptional technologies to efficiently process large quantities of data within tolerable elapsed times.
Technologies include:• massively parallel processing (MPP) databases• data warehouses• Data mining grids• distributed file systems• distributed databases• cloud computing platforms • the Internet, and • scalable storage system
Big Data Solutions
built by
QuerySurge ™
built by
QuerySurge ™
What is ?
• easily deals with complexities of high volume, velocity and variety of data
Hadoop is an open source project that develops software for scalable, distributed computing.
• is a framework for distributed processing of large data sets across clusters of computers using simple programming models.
• scales up from single servers to 1,000’s of machines, each offering local computation and storage.
• detects and handles failures at the application layer
built by
QuerySurge ™
Key Attributes of Hadoop
• Redundant and reliable
• Extremely powerful
• Easy to program distributed apps
• Runs on commodity hardware
Top Vendors
built by
QuerySurge ™
“Spending on Hadoop software and subscriptions will increase to approximately $677 million by the end of 2017, with overall big data market anticipated to reach the $50 billion mark.”
- Wikibon
built by
QuerySurge ™
MapReduce(Task Tracker)
HDFS(Data Node)
Basic Hadoop Architecture
MapReduce – processing part that manages the programming jobs. (a.k.a. Task Tracker)
HDFS (Hadoop Distributed File System) – stores data on the machines. (a.k.a. Data Node)
machine
built by
QuerySurge ™
ClusterAdd more machines for scaling – from 1 to 100 to 1,000
Job Tracker accepts jobs, assigns tasks, identifies failed machines
Name NodeCoordination for HDFS. Inserts and extraction are communicated through the Name Node.
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Task TrackerData Node
Name Node
Job Tracker
Basic Hadoop Architecture (continued)
built by
QuerySurge ™
MapReduce(Task Tracker)
HDFS(Data Node)HiveQLHiveQL
HiveQL
HiveQL
HiveQL
HiveQL
Apache Hive - a data warehouse infrastructure built on top of Hadoop for providing data summarization, query, and analysis.
Hive provides a mechanism to query the data using a SQL-like language called HiveQL that interacts with the HDFS files
• create• insert • update • delete• select
Apache Hive
built by
QuerySurge ™
Data Warehouse Review
about Data Warehouses…Data Warehouse• typically a relational database that is designed for query and analysis rather than
for transaction processing• a place where historical data is stored for archival, analysis & security purposes. • contains either raw or formatted data• combines data from multiple sources:
o saleso salaries o operational data o human resource datao inventory datao web logso social networkso internet text and docso other
built by
QuerySurge ™
Data Warehouse: the ETL process
ETL: Extract, Transform, LoadWhy ETL?Need to load the data warehouse regularly (daily/weekly) so that it can serve its purpose of facilitating business analysis.
Extract - data from one or more OLTP systems and copied into the warehouse
Extract
Transform – removing inconsistencies, assemble to a common format, adding missing fields, summarizing detailed data and deriving new fields to store calculated data.
Transform
Load – map the data and load it into the DW
Load
built by
QuerySurge ™
Data Warehouse: the Marketplace
“The data warehousing market will see a compound annual growth rate of 11.5% …to reach a total of $13.2 billion in revenue.”
- consulting specialist The 451 Group
Data Warehouse sizeSmall data warehouses: < 5 TBMidsize data warehouses: 5 TB - 20 TBLarge data warehouses: >20 TB- Analyst firm Gartner
Leaders in Data Warehouse Data Management Systems
- Analyst firm Gartner’s ‘Magic Quadrant for Data Warehouse Database Management Systems’
built by
QuerySurge ™
Extract
built by
QuerySurge ™
Legacy DB
CRM/ERP DB
Finance DB
Testing the Data Warehouse: the ETL process
Source Data
ETL Process Target Data Warehouse
Transform
Load
built by
QuerySurge ™
Testing Big Data
built by
QuerySurge ™
Data Warehouse & Hadoop:
2 Use Cases:Data
Warehouse
Hadoop
NoSQL
Hadoop Data Warehouse
built by
QuerySurge ™
USE CASE 1*** Use Hadoop as a landing zone for big data & raw data
1) bring all raw, big data into Hadoop
2) perform some pre-processing of this data
3) determine which data goes to Data Warehouse
4) Extract, transform and load (ETL) pertinent data into Data Warehouse
Use Case #1:Data Warehouse & Hadoop
***Source: Vijay Ramaiah, IBM product manager, datanami magazine, June 10, 2013
built by
QuerySurge ™
Recommended functional test strategy: Test every entry point in the system (feeds, databases, internal messaging, front-end transactions).
The goal: provide rapid localization of data issues between points
test entry point
built by
Business Intelligence
software
ETL
Source Data
Source Hadoop ETL Process Target DWH
built by
QuerySurge ™
Use Case #1:Data Warehouse & Hadoop
test entry point test entry points
Use Case #2: MongoDB, Hadoop, DWH &
Relational DB & Data WarehousingSource Data
@
BI, Analytics & ReportingIngestion
built by
QuerySurge ™
QuerySurge ™
test entry point
test entry point
test entry point
test entry point test entry point
built by
QuerySurge ™
Testing Big Data: 3 Big Issues
- we need to verify more data and to do it faster
- we need to automate the testing effort
- We need to be able to test across different platforms
We need a testing tool!
built by
QuerySurge ™
About QuerySurge ™
The Testing Solution
built by
built by
QuerySurge ™
What is QuerySurge ™?
the collaborative Big Data
Testing solution that finds bad data & provides a holistic view of your data’s
health
built by
the QuerySurge advantage
built by
QuerySurge ™
Automate the entire testing cycle Automate kickoff, tests, comparison, auto-emailed results
Create Tests easily with no SQL programming ensures minimal time & effort to create tests / obtain results
Test across different platforms Hadoop, data warehouses, NoSQL, database, flat file, XML
Collaborate with team Data Health dashboard, shared tests & auto-emailed reports
Verify more data & do it quickly verifies up to 100% of all data up to 1,000 x faster
Integrate for Continuous Delivery Integrates with most Build, ETL & QA management software
QuerySurge™ Architecture
Web-based…
Installs on...
Linux
Connects to…
…or any other JDBC compliant data source
built by
QuerySurge ™
QuerySurgeController
QuerySurgeServer
QuerySurgeAgents
Flat Files
built by
QuerySurge ™
QuerySurge™ Modules
Design Library
SchedulingDeep-Dive Reporting
Run Dashboard
Query Wizards
Data Health Dashboard
Fast and Easy. No programming needed.
built by
QuerySurge ™
QuerySurge™ Modules
• Perform 80% of all data tests - no SQL coding needed
• Opens up testing to novices & non-technical team members
• Speeds up testing for skilled SQL coders
• provides a huge Return-On-Investment
Design Library• Create Query Pairs (source & target SQLs)
• Great for team members skilled with SQL
QuerySurge™ Modules
Scheduling Build groups of Query Pairs Schedule Test Runs
built by
QuerySurge ™
Deep-Dive Reporting Examine and automatically
email test results
Run Dashboard View real-time execution Analyze real-time results
QuerySurge™ Modules
built by
QuerySurge ™
built by
QuerySurge ™
Data Health Dashboard• view data reliability & pass rate
• add, move, filter, zoom-in on any data widget & underlying data
• verify build success or failure
QuerySurge™ Modules
(1) Trial in the Cloud of QuerySurgeTM, including self-learning tutorial that works with sample data for 3 days
(2) Downloaded Trial of QuerySurgeTM, including self-learning tutorial with sample data or your data for 15 days
for more information on our Trials, please visit: www.querysurge.com/compare-trial-options
TRIAL IN THE CLOUD
built by
QuerySurge ™
Free Trials & TrainingQuerySurge™
http://www.rttsweb.com/training/courses/big-data-testing-courses
Big Data Testing CoursesFilled with examples and labs, this hands-on training teaches concepts and HQL techniques used in Big Data testing.
For more information on our Big Data Testing classes, please visit:
a last word about Hadoop…
built by
built by
QuerySurge ™
To see the video of this webinar please visit:http://www.querysurge.com/solutions/testing-big-data/big-data-testing-for-hadoop
Big Data and Hadoop are on the verge of revolutionizing enterprise data management architectures.
- DeZyre