Breaking the Kubernetes Kill Chain: Host Path Mount
Flink Case Study: Capital One
1. Flink at Capital One:
Case
Study
Slim Baltagi @SlimBaltagi
Director of Big Data Engineering, Fellow
Capital One
2. 2
Agenda
1. Capital One at a glance
2. Some elements of our Technology
Strategy
3. What is the business problem of this
case study?
4. What is the related solution architecture?
5. What values Flink added to the solution?
3. 3
A leading consumer and
commercial banking institution with
$306.2 billion in assets, $204.0
billion in loans and $210.4 billion in
deposits
– 8th largest bank based on U.S. deposits1
– 4th largest credit card issuer in the U.S.3
– 3rd largest issuer of small business Visas
and MasterCards in the U.S.4
– 3rd largest independent auto loan
originator5
– Largest US direct bank6
Conducts business in the US,
Canada and the U.K.
• More than 65 million customer
accounts and 46,000 associates
• Fortune 500 rank: 124
• Best Companies rank: 85
1. Capital One at a glance
1) Domestic deposits ranking as of Q4’14
2) Source: FDIC, June 2014, deposits capped at $1B per branch
3) Company-reported domestic credit card outstandings, Q1’15, American Express ex Charge Cards
4) Source: Nilson Report, Q4’13
5) Source: JD Power, 2014
6) FDIC, company reports as of Q4’14
4. 4
2. Some elements of our Technology Strategy
Leverage the power of Open Source
technology beyond just a ‘low cost’
alternative.
Introduce new capabilities to address
limitations of our legacy platforms.
Shift the data processing paradigm from a
batch to real-time stream processing.
Build solutions easy portable from on-
premise to the cloud.
Empower our associates to dream, disrupt
and contribute to Open Source projects.
5. 5
3. What is the business problem of this case
study?
Real-Time monitoring of customer activity
data (Audit log event details, failure and
success data, … ) to:
• proactively detect and resolve issue
immediately
• prevent significant customer impact
• enable flawless digital enterprise experience
The current legacy solution uses expensive
and proprietary tools.
The current legacy solution offer very limited
realtime and advanced analytics capabilities.
7. 7
5. What values Flink added to the solution?
A costeffective solution with the same
capabilities as proprietary logdata analytics
tools
Real-Time event processing which was not
possible with our legacy system:
• Reliable realtime, exactlyonce event
processing. Example: Real-Time alerts
• Transformations, enrichments, lookups with
very low overhead in realtime
A future proof solution to handle growing
customer activity data
8. 8
5. What values Flink added to the solution?
More advanced analytics on data streams,
such as:
• Advanced windowing to perform analytics
beyond eventatatime operations.
• Machine learning: Event correlation,
automated fraud detection, event clustering,
anomaly detection, user session analysis, etc
Solution aligned with our technology
strategy
9. 9
Please come to my talk!
Day 1 - October 12, 2015
16:00 - 16:40
Flink and Spark: Similarities and
Differences
@SlimBaltagi
@CapitalOne