Raspberry Pi 5: Challenges and Solutions in Bringing up an OpenGL/Vulkan Driv...
Big Data Business Wins: Real-time Inventory Tracking with Hadoop
1. MetaScale is a subsidiary of
Sears Holdings Corporation
MetaScale is a subsidiary of
Sears Holdings Corporation
Ankur Gupta
General Manager
Big Data Business Wins: Real-Time
Inventory Tracking with Hadoop
2. MetaScale
A big data technology solution
provider and subsidiary of
Sears Holdings that delivers a
full spectrum of services
focused on big data and
Hadoop
MetaScale is a ‘big data
accelerator’. We provide the
technology, talent, and
solutions to accelerate the
value from big data
Offices in Chicago, San Jose,
and Pune, India
A Fortune 100 company, nearly $40
billion in annual revenue
The nation’s fourth largest broad
line retailer with almost 2,500 full-
line and specialty retail stores in the
US and Canada
A front runner in big data efforts
including driving personalized
marketing and generating savings
from legacy migration
Running one of the biggest rewards
programs that captures and
analyzes very large number of
customer transactions quickly
Our Parent Company
Sears Holdings
2
3. What tools can be used to migrate Point-of-Sales
(POS) data from different legacy systems to Hadoop
Establishing an Enterprise Data Hub with Hadoop in
order to create a single version of truth
What is a reference architecture for near real-time
inventory tracking
3
Objectives
4. From a recent Wikibon survey:
Enterprise practitioners believe the potential value of
Big Data is significant
However, many are struggling to derive maximum
value from their big data investments
• 46% of Big Data practitioners report that they have only
realized partial value from their Big Data deployments
• 2% declared their Big Data deployments total failures, with no
value achieved
Challenge of Achieving Big Data ROI
Source: Enterprises Struggling to Derive Maximum Value from Big Data, Wikibon, Sep 2013
http://wikibon.org/wiki/v/Enterprises_Struggling_to_Derive_Maximum_Value_from_Big_Data
4
5. According to Wikibon, three compelling reasons for this
struggle to achieve maximum business value from big
data…
1. A lack of skilled Big Data practitioners
2. "Raw" and relatively immature technology
3. A lack of compelling business use case
Challenge of Achieving Big Data ROI
Source: Enterprises Struggling to Derive Maximum Value from Big Data, Wikibon, Sep 2013
http://wikibon.org/wiki/v/Enterprises_Struggling_to_Derive_Maximum_Value_from_Big_Data
5
6. Making Business Decisions Quickly
6
The Hadoop ecosystem gives
business the ability to create value
from its data by being able to process
and store vast amounts of data from
disparate sources.
Hadoop enables faster processing on larger data sets
for analytics and deep analytics.
Storm, Kafka and Cassandra provide the technology for
real-time analytics to make business agile.
7. Keys for Achieving Big Data Success
7
Bring IT and Business together
Define realistic success criteria
Ask “what are you really trying to accomplish?”
Understand how Hadoop will fit into your environment
See the end results first before you start your journey
Discover your big data use case!
9. Real-Time Analytics with Cassandra
By implementing Hadoop and Cassandra into a
traditional environment, Business Intelligence teams
are able to provide more accurate and real-time
inventory, pricing, sales and return data as well as
predicting ideal floor plans.
Managing inventory with up-to-the-second data...
9
In-Store
Purchases
Online
Purchases
Real-time
inventory data
ensures that
items ordered
are in-stock.
10. POS data was stored in different formats in different
legacy systems (Mainframe and Teradata)
No single version of truth
No real-time capability
Inventory
Batch File Sent
ONCE A DAY
CHALLENGE
This latency resulted in potential loss of sales and customer
dissatisfaction when items are ordered that are no longer in stock.
10
Real-Time Analytics with Cassandra
POS Volume
Average 100,000 message per day
Peak 77,000 messages in 1 hour at
4:00am the day after Thanksgiving
11. SOLUTION – Phase 1
Condense all POS data from different legacy
systems and applications into Hadoop
Enterprise Data Hub
Create a Single Version of Truth
11
Real-Time Analytics with Cassandra
Hadoop enables a single version of truth for deep analytics,
but there is still no real-time capability…
12. SOLUTION – Phase 2
12
Real-Time Analytics with Cassandra
Use Kafka to extract messages from
POS queue
Kafka sends messages to Cassandra
for real-time processing
13. SOLUTION – End-to-End
Messages are sent from Cassandra to
Hadoop for back-end, deep analytics.
13
Real-Time Analytics with Cassandra
4 Node 4 Node
11 Node
14. Faster decision making…
Business Intelligence Teams
are able to provide more
accurate and real-time
inventory, pricing, sales and
return data.
BEFORE Cassandra
Real-Time Solution:
Inventory Batch File
Sent Once a Day
Real-Time Analytics with Cassandra
AFTER Cassandra
Real-Time Solution:
Inventory Data Sent
in Sub-Milliseconds
14
RESULT
15. Increased sales by improving item
availability.
Real-Time Analytics with Cassandra
15
Value for the Organization
Increased customer satisfaction
because customer is able to get
what was ordered.
16. Real-Time Analytics with Cassandra
16
Value for the Organization
Cost savings from reduced
customer service center calls.
Aha Moments
Cost savings from reduced truck
load times.
18. Hadoop Enterprise Data Hub gives business users access to
more data from more sources for deep analytics.
Hadoop Enterprise Data Hub
18
Single Version of Truth
19. Firewall Issues
Normally, Storm or Kafka can be
used to send POS messages to
Cassandra.
In certain situations where a firewall
exists between data source and
processing cluster - such as created
by mergers or spin-outs – both
Storm and Kafka can be used to
send messages over the firewall.
19
Unique Challenge for a Complex Enterprise
21. Advanced Analytics
21
Inventory forecasting with
Machine Learning on data from
Weather Reports
Data-Driven Decision Making
Once the Hadoop / Cassandra framework is in place, data
from virtually any source can be consumed in the Enterprise
Data Hub for Advanced Analytics.
New ways to use Social, Geo,
Sensor data to develop
predictive models…
23. Enterprise Data Hub and single version of truth for all data
Hadoop can help you answer questions that were difficult
or cost prohibitive to answer before
Hadoop can transform your organization’s approach to
how you use data and ask questions you never even
thought of
Must have a clear strategy and
long-term plan
Leverage the right partnerships to
achieve your goals
23
Big Data Business Wins