Grid computing in the cloud for Financial Services industry - CMP205-I - New York AWS Summit

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.S U M M I T
Grid computing in the cloud
for Financial Services industry
Innovating without infrastructure constraints
C M P 2 0 5 - I
Barry Bolding
Director, WW HPC EC2 BD
AWS

Financial modeling has grown more onerous
Compute-intensive calculations
More granular risk factors
Wider range of scenarios
More historical data
Broad regulatory requirements
Comprehensive Capital Analysis and
Review (Banking/Dodd Frank)
Solvency Capital Requirements
(Insurance/Basel II)
Fundamental Review of the
Trading Book (Insurance/Basel III)
Diverse risk analysis models
Market risk
Credit risk
Liquidity risk

And on-premises grids are often inadequate
Data center capacity is limited, resulting
in simulation backlogs or inadequate
risk calculations
Financial instruments require flexible
compute resources for development
and testing.
Limited capacity, which results in long
runtimes for simulations
Regulatory and market fluctuations
require flexible compute capabilities
Large upfront investments and
maintenance required to run
on-premises grids
Standardized hardware offers limited
grid and compute types

And the choices poor
$
Optimize for cost
Utilization is high but users may
wait a long time to access the system
Optimize for availability
Usage remains low which
drives up costs

Matching demand to capacity is challenging
Attempting to match variable computing demand to static on-premises compute grids is
extremely difficult, and adding capacity is time- and capital-intensive
Time
Capacity
Actual demand for computing
Server
acquisition
Server
acquisition
Server
acquisition
Project
delay
Project
delay
Project
delay
Demand exceeds
capacity
Unused
IT resources

What if you could add 100,000 cores of your choice to your compute grid as you needed
them, remove them when you are done, and only pay what you used?
The solution: On-demand elastic compute power

What does the solution look like?
25,000
50,000
75,000
# of Cores
0
Time: +00 hrs
Scale using Elastic Capacity
<1,000 cores

25,000
50,000
75,000
# of Cores
0
Time: +24 hrs
>75,000 memory-optimized cores

25,000
50,000
75,000
# of Cores
0
Time: +72 hrs
<1,000 cores

25,000
50,000
75,000
# of Cores
0
Time: +120 hrs
>30,000 GPU optimized cores

Cloud is increasingly becoming an
attractive option for Grid Computing
Why are Financial Services institutions increasingly shifting
grid computing to the cloud?
Virtually unlimited
compute
and storage
resources
Various
instance types
for specific types of
workloads
Cost
optimization
with a variety
of pricing structures
Enhanced
security
and compliance
Expanded
big data
capabilities
for analysis
and business
intelligence
Automation
capabilities
for scaling and
provisioning
resources
Management
consoles
to provide
enterprise-wide
control and
visibility

Transform your
risk practices
Compute capacity as catalyst
for modernizing risk functions
Actuarial productivity
Improve risk insight
Optimize reserving levels
Respond to regulatory changes
Store new models &
output files in the cloud
Support for running models more often;
storage for increasingly complex and larger
models and
more numerous output files
Consolidated single source of truth
Move HPC grid to cloud
CAPEX to OPEX
Grid modernization
Variable use to an on-demand,
elastic utility
Augment grid, or replace grid
Moving HPC grids to AWS provides burst capacity to actuaries and data scientists leading to
faster model runtimes, enabling more analysis, and improving risk insights
Cloud-enabled grids offer cost and business benefits

These are the most popular grid computing use cases for
the cloud
CCAR/FRTB & other
risk-based capital
regimes
Trading and Pricing
models Valuation Hedging strategies Backtesting

Why Grid Computing or HPC on AWS?
Virtually unlimited infrastructure enabling
scaling and agility not attainable on-premises
Flexible configuration options quickly iterate resource
selection and ensure cost optimization
Instant access to latest technologies with no lengthy
procurement cycles or big capital investments
Better ROI
Faster time
to results

Global Infrastructure
We add the equivalent of an entire Fortune 500 company’s compute capacity every day
Coming soon
66 Availability Zones
within 21 geographic
Regions around the world

High Performance Computing on AWS

AWS Services to get started with HPC on AWS
AWS Budgets
NICE DCV
Amazon AppStream 2.0
Visualization
AWS Batch
AWS ParallelCluster
NICE EnginFrame
Automation &
orchestration
Amazon EBS
with Provisioned IOPS
Amazon FSx for Lustre
Amazon EFS
Amazon S3
Storage
Amazon EC2 instances
(CPU, GPU, FPGA)
Amazon EC2 Spot Instances
AWS Auto Scaling
Placement groups
Enhanced networking
Elastic Fabric Adapter
Compute &
networking
AWS DataSync
AWS Snowball
AWS Snowmobile
AWS Direct Connect
Data management
& data transfer
Amazon CloudWatch
AWS Identity and Access Management (IAM)

Broad HPC partner community
Application partners
Infrastructure
partners
Technology
partners
Consulting
partners

HPC workloads across industries
Life Sciences Financial Services Oil & Gas
Design & Engineering Climate & Geosciences Autonomous Vehicles

HPC workloads with different compute
and throughput characteristics
Tightly-coupled workloads Loosely-coupled workloads Accelerated computing
Visualization AI/ML High volume data analytics

High Performance Computing (HPC) on AWS
Virtual Private Cloud on AWS
3D graphics virtual workstation
License managers and cluster head nodes
with job schedulers
Cloud-based, automatic scaling HPC clusters
Shared file storage Storage cache
On AWS, secure and well-
optimized HPC clusters can
be automatically created,
operated, and torn down
in just minutes
Amazon S3
and Amazon S3 Glacier
On-premises
HPC resources
Corporate data center
AWS Snowball
AWS Direct Connect
Thin or zero client—
no local data
Third-party IP providers
and collaborators
Machine learning
and analytics

Simple steps to get started

Broadest and deepest platform choice
Categories Capabilities Options
(AWS, Intel, AMD)
(up to 4.0 GHz)
(up to 12 TiB)
(HDD and NVMe)
(up to 100 Gbps)
(GPUs and FPGA)
(Nano to 32xlarge)
+ + =
200+
instance types
NEW
NEW
NEW

Amazon FSx for Lustre: High and scalable performance
Each terabyte (TB) of storage provides 200 MB/second of file system throughput and ~5,000 IOPS
High and scalable
performance
Parallel File System
100+ GiB/s throughput
Millions of IOPS
Consistent sub-millisecond latencies
Supports concurrent access
from hundreds of thousands
of cores
SSD-based

AWS Batch
AWS Batch dynamically provisions resources, plans, schedules, and executes
No additional components to install
Event
Changes in
data state
Requests
to endpoints
Services (anything)
Scheduled
triggers
Compute
Execution
Your code
Auto Scaling
Job queue

Easy cluster management: AWS ParallelCluster
Simplifies deployment of HPC in the
cloud, including integrating with
popular HPC schedulers
Integrated with AWS Batch, Amazon
FSx for Lustre and
Elastic Fabric Adapter
Link to Tutorial <INSERT>

“
”
We can be far more secure in the cloud and achieve a higher level of assurance at a much lower cost, in terms of effort and
dollars invested. We determined that security in AWS is superior to our on-premises data center across several dimensions,
including patching, encryption, auditing and logging, entitlements, and compliance.
– John Brady, CISO, FINRA
AWS is the first choice for highly regulated organizations
Security enhancements from
1M+ customer experiences
AWS industry-leading
security teams: 24/7,
365 days a year
Security infrastructure
built to satisfy military,
global banks, and other
high-sensitivity
organizations
Over 50 global
compliance certifications
and accreditations

Simplifying capacity and cost optimization
Use Reserved Instances for
known/steady-state workloads
Scale using Spot Instances, On-Demand
Instances, or both
Evaluate the trade-off of time
to solution vs. cost for scaling

Innovations in HPC infrastructure
Up to 4 GHz sustained, all-turbo performance
• Z1d instances are optimized for memory-intensive, compute-
intensive applications
• Custom Intel Xeon Scalable processor
• Up to 4 GHz sustained, all-turbo performance
• Up to 385 GiB DDR4 memory
• Enhanced networking, up to 25-GB throughput
HPC stack on AWS
License managers and cluster head
nodes with job schedulers
Cloud-based, auto-scaling HPC clusters
Featuring

Innovations in HPC infrastructure
Massively scalable performance
• C5n Instances will offer up to 100 Gbps of network
bandwidth
• Significant improvements in maximum bandwidth, packet
per seconds, and packets processing
• Custom designed Nitro network cards
• Purpose-built to run network bound workloads including
distributed cluster and database workloads, HPC, real-
time communications, and video streaming
HPC stack on AWS
License managers and cluster head
nodes with job schedulers
Cloud-based, auto-scaling HPC clusters
Featuring

Manage 50X the number of securities
4,000 times faster
In hours, instead of months
Run risk models
Helping financial institutions
model investment risks
S U M M I T

Bankinter: Intraday Credit Risk Simulation
Javier Roldán
Director of Technological Innovation
”
“ The challenge
The bank needs to run 5 million credit risk simulations to evaluate the
financial health of Bankinter’s clients.
The solution
Implementing simulations in parallel on a grid of Amazon EC2 instances to
obtain the result in a very short time period.
The result
The bank has brought down the average time for running simulations from
23 hours to 20 minutes and estimates it would spend 100 times more in
hardware alone if it chose to exit the cloud.
With AWS, we now have the power to decide
how fast we want to obtain simulation results.
More important, we have the ability to run
simulations that were not possible before due
to the large amount of infrastructure required.
98% decrease in calculation time
and savings of 100x vs. on-premises
Key benefits

Talanx: Insurance Solvency II Model Simulation
Achim Heidebrecht
Head of group IT at Talanx
”
“ The challenge
Producing precise, high-quality, and time-sensitive Solvency Capital
Requirements and Minimum Capital Requirements using a complex Monte
Carlo simulation for the quarterly and annual company reports without
creating bottle-neck in internal IT.
The solution
With AWS, 4 distinct environments set up in 4 Virtual Private Clouds (VPCs)
using at peak 550 Amazon EC2 instances and 280 TB storage.
The result
75% reduction in calculation time and about 8 million
euros in annual savings.
Using AWS we are already seeing a
75% reduction in calculation time, and €8m in
annual savings, when running
our Solvency II simulations, while
still complying with our very strict
data policies.
75% reduction in calculation time
and €8 million in annual savings
Key benefits

Accenture and AWS help FIs calculate risk
“ Accenture’s Risk Calculation for cloud solution (ARC) provides a
framework that enables financial institutions to adopt AWS for
compute-intensive risk calculation use cases while helping control
costs and retain full data security and confidentiality
Some key use cases include:
• Fundamental Review of the Trading Book Internal Model Approach (FRTB
IMA) calculations—Value at Risk (VaR-ES) and Standardized Approach (SA)
• Increased stress-testing demands through both increased range of
scenarios run and the frequency of calculation
• Periodic calibration and validation of calculations—that is, selection of
market risk stress period, profit and loss (P&L) attribution
Accenture and AWS help
financial organizations worldwide
stay compliant through cloud agility, improved
enterprise architectures,
and transparent controls which support much
needed computation headroom (capacity and
capability) and agility
Enhanced ability to meet regulatory
requirements while reducing costs

TIBCO’s GridServer solution runs on AWS
Nelson Petracek
Global Chief Technology Officer, TIBCO
“ By running TIBCO GridServer on AWS, financial institutions can take advantage
of on-demand cloud computing with an almost infinite capacity for
computation-intensive applications
TIBCO’s solution can analyze large volumes of data in connection with risk,
trading, financial models, and more. For example, the GridServer software can
turn risk reports for stock traders from a six-to-eight-hour process to a 15-
minute intraday cycle, enabling traders to make more educated decisions
GridServer will also supplement investment banks’ capacity for daily
Fundamental Review of the Trading Book (FRTB) calculations, which will
significantly increase the need for large-scale grid computing environments
when the requirement goes into effect in 2019
With the TIBCO Connected Intelligence platform
of solutions, companies can build a system that
automatically allocates cloud resources on
demand. GridServer embeds an AWS
architecture that is highly scalable, both in
terms of speed and throughput, and is ideal for
industries that require high-volume
calculations, such as financial services,
healthcare, and oil and gas.
Accelerated go-to-market timelines
for products and services

What would you do with 1 million+ vCPUs?

MAPFRE: Insurance Solvency II solvency check
Guillermo Diez Perellón
Director of IT Architecture
for MAPFRE
”
“ The challenge
The company is required to perform a monthly solvency check
to test its risk under worst-case scenarios. Running these
calculations requires high performance computing (HPC)
machines that are used only a few times a month.
The solution
To gain compute capacity while maintaining data privacy,
the company established an Amazon VPC to move its data
to Provisioned IOPS Amazon EBS and share the data to the
Amazon EC2 cluster.
The result
By using Amazon EC2 the company achieved cost savings of €820,000 over
three years compared with its on-premises solution.
We asked AWS to increase the capacity
and number of sessions and help us with
the configuration. Now, the system is
working very well.
€820,000 in savings over 3 years
Key benefits

Aon Benfield: Actuarial risk modeling & hedging
Peter Phillips
Managing Director,
Aon Benfield Securities
“ The challenge
When Aon first launched its financial modeling tool, PathWise, it
used a broadband HPC processor in a colocated data center, but found it
needed a more scalable service than the colocation facility could provide.
The solution
To gain the scalability and cost savings it needed, Aon moved its infrastructure
to AWS and depreciated its colocated data center.
Aon built a front end on AWS for its processing solution, automatically running
GPU instances on Amazon EC2 using Amazon EBS in an Amazon VPC for
security.
The result
By moving its infrastructure to AWS, Aon became 500 times more cost efficient
for its clients, and reduce a 10-day process to 10 minutes.
Using AWS helps us reduce a 10-day
process to 10 minutes. That’s
transformative: it broadens our ability to
discover.
€820,000 in savings over 3 years
Key benefits
”

Compliance programs
SOC 1 SOC 2 SOC 3
Global
Asia
Pacific
Europe
United
States

2
2 2
4
2
1
1
3
7
7
4
9
5
7
6 6
7
7
4
8
4
Cores
8
2
1
9
5
4
5
3
1
2
3
6
1
9
4
8
1
2
8
7
7
6
Fixed Data Center
Capacity Limit
Cores
Finite capacity, usually with
long queues to wait in
Massive capacity when needed to speed up time to results, and agile
environment when additional hardware and software
experimentation is needed
Remember: The metric for success for any business should
be time-to-results

Remember: A TCO analysis never tells the whole story
72.8%
Lost productivity & longer time to results
of organizations that use HPC reported delayed or
cancelled HPC jobs*
Lost innovation
Questions are left unasked,
experiments are left undone, and
potential revenue
left on the table.
Outdated technology
Almost 20% of the useful
life of new technology/ hardware
lost in the procurement process.
Technical debt
Adapting newer algorithms to meet
the requirements of an existing
infrastructure = delays, and below-
par performance.

Flexible configuration and virtually unlimited scalability
to grow and shrink your infrastructure as your HPC workloads
dictate, not the other way around
HPC on AWS

Thank you!

Please complete the session survey in the mobile app.
1. Tap the Catalog icon.
2. Filter by Show Past Sessions.
3. Select the session that you attended.
4. Tap Complete an Evaluation to submit your feedback.
Complete three surveys, and you’ll receive a gift at the Help Desk.

Compute needs in financial services are changing
Financial simulations are essential to the operations of all financial institutions to
identify and manage risk, optimize capital, and make informed investment and
pricing decisions
The development of new products and trading strategies, particularly for
complex products, require a greater variety of more complex, and more
frequently backtested, datasets
Regulations require financial institutions to perform stress testing, while
regulatory changes have increased the complexity of allocating capital and
collateral to meet margin and solvency requirements

Grid reference architecture
Subnet Placement Group
MSS
Node
Scheduler
Node
Compute
Nodes
Compute
Nodes
Metadata
Servers
Data Node
Servers
Amazon S3
IAM roles
Amazon EFS
Virtual Private Cloud
AWS
Batch
Amazon CloudWatch AWS
CloudTrail
AWS
Config
AWS
KMS
AWS
IAM
AWS
CloudFormation
AWS
Snowball
AWS Direct Connect
Endpoints

Risk models in hours vs. months

Flexible, cost effective migrations for Financial Services
Easier migrations
Flexible deployment modes
Stretch cluster mode
Multi-cluster mode
Spot Enablement

Grid operations
Corporate data center AWS Cloud
Amazon S3

Pacific Life Insurance: Overcoming hardware constraints
John Trujillo
Assistant VP of Technology,
Pacific Life
“ The Challenge
In order to increase the speed of actuarial workloads used to set pricing and
create new products, Pacific Life needed to be able to scale up their high
performance computing capabilities on demand.
The Solution
Pacific Life turned to AWS as part of a hybrid computing environment, using
the AWS Cloud in combination with data centers.
The Result
By using AWS, the company can quickly scale up additional compute capacity in
minutes with lower costs and reduced IT overhead compared with adding to its
own data center assets. The company also benefits from the AWS robust
security disaster recovery protocols, enabling them to focus more on
innovation and experimentation.
AWS helps us to experiment more than
we otherwise would have…In the long-
term I definitely see us relying on AWS as
a provider for absolutely critical business
services.
Accelerated go-to-market timelines
for products and services
”

Running HPC applications
at extreme scale
“Storage technology is amazingly complex and we’re constantly pushing the limits of
physics and engineering to deliver next-generation capacities and technical innovation.
This successful collaboration with AWS shows the extreme scale, power, and agility of
cloud-based HPC to help us run complex simulations for future storage architecture
analysis and materials science explorations. Using AWS to easily shrink simulation time
from 20 days to 8 hours allows Western Digital R&D teams to explore new designs and
innovations at a pace unimaginable just a short time ago.” – Steve Phillpott, CIO, Western
Digital
single
HPC cluster of 1 million vCPUs
Accelerating time to innovation
20 days → 8 hours

Takeaways
Compute-optimized instance innovation keeps driving price/performance lower
Instance differentiation drives Grid ramp-up speed, availability at scale
Native HPC storage makes easier, faster, less expensive to run grids
Broad offering of HPC solutions from AWS and our partner network

Fannie Mae: Mortgage modeling
The challenge
Fannie Mae’s seven-year-old HPC grids
required more than six months to add
incremental compute capacity,
provided only limited I/O capacity and
storage, and was accessed through a
complex API, impacting their ability to
develop new applications.
The solution
Fannie Mae began to work with AWS
using AWS Lambda to build the first
serverless HPC computing platform in
the industry and Fannie Mae’s first
program AWS Cloud native
application.
The result
By March 2017, Fannie Mae had successfully deployed their first financial modeling application
to preproduction using 15,000 concurrent Lambda instances. Using AWS allows Fannie Mae to
run a simulation of 20 million mortgages in only two hours, more than three times faster than
the previous process.

Large Global Insurer: Stress-test calculations
The challenge
A large global insurer was using their internal
compute grid to calculate the valuation and
risks associated with its portfolio of variable
annuities. The company was also required to
participate in the Federal Reserve’s annual
Comprehensive Capital Analysis and Review
(CCAR) stress test, requiring it to scale its
internal grid by a factor of four in order to not
impact existing workloads.
The solution
Working with APN partners Cycle
Computing and 2nd Watch, the
company used AWS
to run its CCAR calculations
completely in the cloud.
The result
The CCAR program was so successful the company started migrating its existing variable annuity jobs to
AWS and is now running its grid entirely in the cloud. The company’s $40 billion book now runs overnight
on 8,000 cores. The migration to AWS allowed the company to reduce its operational costs and transform
the way it approaches its business. Without the delays associated with its on-premises grid, it has been
running more complete analytics, allowing it to accelerate its product development and risk activity.

Large Hedge Fund: Quantitative research
The challenge
A quantitative trading firm wanted to
provide additional modeling capacity to its
researchers responsible for creating and
developing risk and trading algorithms in
order to reduce the time required to see
the results of their models and allow them
to more rapidly improve the efficacy of
their trading strategies.
The solution
The firm piloted an implementation of its
existing grid infrastructure, which could
utilize Amazon EC2 On-Demand Instances,
significantly reducing analysis time.
Reluctant to increase costs, not wanting to
lose existing responsiveness to the users,
and not beholden to any regulatory or
reporting deadlines, the firm turned to
Amazon EC2 spot pricing.
The result
Today, the firm is able to deploy over 80% of its instances using spot pricing, and has achieved a
75% reduction in costs compared to the use of On-Demand Instances. This grid is deployed
across multiple Availability Zones in the eastern US and can peak up to 75,000 cores. The
availability of accelerated compute instances was also a key benefit as their grid relies primarily
on GPU-enabled instances.

Grid computing in the cloud for Financial Services industry - CMP205-I - New York AWS Summit

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Was ist angesagt?

Was ist angesagt? (20)

Ähnlich wie Grid computing in the cloud for Financial Services industry - CMP205-I - New York AWS Summit

Ähnlich wie Grid computing in the cloud for Financial Services industry - CMP205-I - New York AWS Summit (20)

Mehr von Amazon Web Services

Mehr von Amazon Web Services (20)

Grid computing in the cloud for Financial Services industry - CMP205-I - New York AWS Summit