The automotive industry is undergoing rapid change in every area, and cloud computing is at the center of this transformation. In this session, learn how leading global automotive companies are partnering with AWS to transition their enterprises to create innovative, connected brand experiences and future mobility services. Volkswagen Group discusses how they are migrating legacy applications to create an agile, scalable, microservices-based digital platform using AWS as a cloud service provider, serving their 12 global brands and connecting millions of vehicles. BMW Group shows how they modernized their digital customer engagement platform by migrating their monolithic backend to a microservice-based platform and subsequently leveraging multiple AWS services to support more than 28 front-end customer touchpoints across auto and motorcycle brands. Finally, Toyota Research Institute shows how they are using deep learning on Amazon EC2 P3 instances, Amazon S3, Amazon SQS, and AWS networking services to accelerate the research and development of their automated driving systems.
32. Page 32BMW Group | AWS re:Invent 2018
THE PLATFORM CONSOLIDATES THE VEHICLE PRODUCTS AS SERVICES.
~7
Engines
~12
Colors
~14
Wheels
~11
Upholstery
~7
Trims
~90
Optional
Equipment
many possibilitiesto configure your dream car
=
OUR PLATFORM CONSOLIDATES THE VEHICLE PRODUCTS AS SERVICES.
33. Page 33BMW Group | AWS re:Invent 2018
EXAMPLES OF FRONTENDS WHICH USE OUR CONFIGURATOR PLATFORM.
Dealer iPad Configurator bmw.de mini.at bmw-motorrad.de 26+ Systems
34. Page 34BMW Group | AWS re:Invent 2018
UNIFIED CONFIGURATOR PLATFORM: CENTRAL
ENGAGEMENT PLATFORM AS BUSINESS ENABLER.
Unified Configurator Platform
Accessories
Product Data
Officialtechnical
Data
Net-Gross-Price
Calculation
Virtual Garage
Constructability
Check
Data Import Data Generation
Product Data Technical Data …
consumes API
...
Technical
Calculations
Key facts:
- RESTful API
- used by ~30 products
- hundreds of Mio API
calls per month
- supports all BMW Group
brands and markets
36. Page 36BMW Group | AWS re:Invent 2018
A MONOLITHIC SYSTEM ISN‘T BYDEFINITION A BAD THING.
Monolith
Team
clients
Benefits:
Fast development of a first production ready MVP
No need to handle the challenges of a distributed system
Challenges and driving factors for going the next step:
Increased number of requirements led to an increase of
the overall complexity
37. Page 37BMW Group | AWS re:Invent 2018
MICROSERVICES WITH API-FIRST STRATEGY: ENABLES FASTER DEVELOPMENT AND
INCREASED TRANSPARANCY.
Benefits:
Increased transparency for API consumer due to API
first approach
Multiple teams can work independently
Challenges and driving factors for going the next
step:
Latency performance to provide our service on a world
wide basis
Broader solution scope needed to satisfy business
needs
Teams
clients
MS
MS
MS
MS
40. Page 40BMW Group | AWS re:Invent 2018
GAME CHANGER: BRINGING THE MICROSERVICES IN A „LIFT-THINK-SHIFT“-APPROACH
TO AWS.
AWS Elastic Beanstalk
AWS API-Gateway
AWS ECS + AWS Batch
AWS RDS
AWS-Cloud
Glassfish 4
Microservice
Software Business Logic
API Specification Transformation into API-GW
Transformation from
ORACLEto Postgres
API-Sec. Auth. Information
Dockerize the business logic
Specifications
WEB-EAM Master Solution
On-Premise Off-Premise
API-Interface (generated)
Authentication Information
Docker + GF4
Microservice DB
Software Business Logic
Rewrite import procedures
from PL/SQLto Java
Microservice DB
Import Job
Docker
Software Import Job
42. Page 42BMW Group | AWS re:Invent 2018
MOVINGTO AWS ENABLED US TO INCREASE OUR IN-HOUSE EFFORT.
DEV OPS
Code Deploy
Test Monitor
DEV OPS
Code Deploy
Test Monitor
BEFORE NOW
43. Page 43BMW Group | AWS re:Invent 2018
TRANSITIONING TO AWS WITH 100% IN-HOUSE EFFORT PAID OFF.
- Reduced timeto market – daily releases instead of 3 major releases per year.
- Supports agile culture for flexible and continuous improvements.
- Enables building solutions on state of the art technologies.
- Enables new business opportunities.
44. LEARN MORE
Visit the re:Invent website to view our session from yesterday
AMT305 – Building BMW Group’s Customer Engagement Platform on AWS
Highlight how AWS provides the resources and capabilities to enable companies of any size and in any industry to innovate and change to be digital innovators
We know that the industry is undergoing a transformation unlike anything it has faced in the past.
Driven by mega trends like Ubiquitous Connectivity, Cloud Computing, Powertrain Electrification, and the Sharing Economy
But also a shift in consumer mindset – they are eschewing traditional automotive purchase reasons and focus more on experiences, and personalized interactions and the mobile revolution has created expectations that their digital lives should follow them wherever they go
For the auto industry, the convergence of these factors is unlike anything we have experienced in the past – these are changes to the fundamental way the customers purchase and drive cars and forcing the entire industry to rethink their brands and business.
https://venturebeat.com/2017/03/22/why-ai-will-drive-the-future-of-connected-cars-starting-now/
In short - Every automotive company will be a technology company
This is a major change to a century-old industry who’s primary focus has been on the hardware: QRD, Fuel Economy, Performance
Now in addition to developing compelling, safe and efficient products, you also have to have a software focus both inside and outside of the vehicle
Now, a customer will choose a vehicle based on weather it has Car Play or not
Our role in the value chain is as an enabler – and we do this in 4 primary ways
Our core IT infrastructure services of Compute, Storage, Networking, Data Base and many others –
Next is a range of digital products and solutions that provide the capability to deliver advanced functionality to our customers
We take this vast toolset that we have and develop reference architectures that allow customers to get started using the most secure, efficient and powerful architecture
And finally, by working with other Amazon Business units and a wide variety of partners to develop automotive solutions that are market ready
When we approach the business, our focus is on 6 core areas. The primary focus is on product solutions around Connected Vehicles, Autonomous Driving Development and Digital Customer Engagement. This is where the marketing is focused and you will see this reflected in our AWS for Automotive web site.
But we also are working through our account teams on the enterprise solutions, such as applications for big data and analytics for predictive maintenance, and marketing applications; High Performance Computing for the R&D divisions and other Enterprise workloads like ERP, SAP, website platforms and call centers.
Exlain product solutions means….
Connected Vehicle Services Platform
Connected Consumer Initiatives – Mobile, B2C, Digital Marketing
Location-based Services
Alexa Voice Integration
In-vehicle App Store and Payments
Amazon Music, Audible, Prime Video
Logistics and Prime Now
Autonomous Vehicle Driving Development
Machine Learning - Predictive Diagnostics
Machine Learning – Targeted Offers
So today we are delighted to be joined by three of our customers who are leading the digital transformation in the automotive industry to tell their stories of how AWS helps them. What is interesting, is that even though each company will be talking about a different area of transformation, you will hear similar themes of how AWS has enabled them to transform their entire organization to be more agile, innovative and responsive to the business and ultimately, their end customer.
To get us started, I would like to introduce Peter Garzarella, head of SW development at Volkswagen Group
Wrap up by inviting others to attend your session
I would like to take you on to our journey on how we modernized our digital customer engagement platform by migrating our monolithic backend to a microservice-based platform and subsequently leveraging multiple AWS services.
- Let us start by talking about a common use case:
Can you image how many possibilties you have to configure your dream car?
For example for the brand new BMW 3 series you have the following possibilities to configure your dream car. You can choose from:
So from a frontend perspektiv you need on the one hand all the vehicle data to visualize it for the customer and on the other hand functionalities like a constructability check, to check if the dream car configuration of our customer is constructable.
Within the BMW Group there existed a huge demande to get the car informations and functionalities from a central platform
So that not everyone needs to take care about where to get the vehicle data from and to rebuilding commonly used functionalities
Therefor we created the unified Configurator Platform that is currently used by more then 30 Applications/Frontends.
As an example our Platform is used from
various car configurators that are for example implemented as nativ iOS Applications or Web Frontends
But our platform is also used for example for getting relevant data for AI purposes or for creating invoices
Let us now have a look on our platform from a high level perspective
We are providing a RESTFul API that exposes all necessary product data and functionalities like
The constructability check / Price calucation services / or to save a customer car configuration within a virtual garage
To expose the data we are frequently importing data from the BMW Group backend systems
Our platform surves currently hundreds of Mio API calls per month on a world wide basis
And our platform is supporting all BMW Group brands and all markets
I would like to talk now about our journey from a monolith system to microservices
And followed that about the important strategic change we did and the transition to AWS
It‘s nowadays modern and everyone is talking about microservices.
But from our view a monolithic system has also advantages:
We started at the beginning with a monolithic system that allowed us to have a fast development of a first production ready MVP
And we do not had the need to handle the challenges of a distributed system just at the beginning like:
Data Synchronization / Debugging and logging and so on…
< CLICK >
Nevertheless as the monolithic system growed we were faced with the challenges a monolith provide:
Small changes led to a rebuild of the complete application
Implementing new requirments led to a lot of refactoring
In a nutshell: The monolith became to complex and the Implementation became slow and we couldn‘t hold on to the development speed that we had at the beginning – So it was time to break up the monolith!
At the same time as we broke up the monolith into microservice to handle the complexability, we also introduced the API first approach.
The API first approach helps us to get all our stakeholders on the same page and to specify the API upfront before the actual implementation starts
<CLICK>
As our platform became more and more popular - new challenges came allong:
How can we provide our platform on a world wide basis by having low latencies?
How can we get a broader solution scope to satisfy business needs with state of the art technologies?
In a nutshell: The microservices helped us to handle the complexity but it was now time to bring the microservices to next level:
But before bringing the Microservices to next level we did in our oppinion an important strategic change!
By Looking from a historical perspektiv: Many companies tried to reduce costs by outsourcing their IT.
But nowadays IT becomes more and more an enabler and a strategic asset to have inhouse
Therfor we decided to getting the IT back inhouse.
In our oppinon together with the services and automation posibilities that the cloud provides today, it‘s worthwhile to evaluate if the transformation from on-prem to off-prem is feasable with inhouse effort.
We did so and it was the best decision we could made!
Let’s see how we did the transition to AWS:
<go step by step through the layers>
- As you can se for us it was important to have as many managed services as possible so that we can benefit 100% from the potential the cloud provides today and that we can focus on our main mission to provide new platform services for the BMW Group.
Before doing the transition to AWS in 100% in-house effort:
Our work was mainly driven from an architectural perspectiv, we described how the architecture should look like and defined the software to build
Now we are taking on more on more phases from the DEV/OPS-Cycle:
We take care of the complete CI/CD Pipeline
We develop more and more of the software components inhouse
We are doing most of the operation tasks inhouse with providing 24/7 on call
So We are coming more and more in the situation of „you build / you run it“
The strategic step of getting back the IT was for now the absolut right decision.
It helps to build up sustainable knowledge within the bmw group and last but not least it also led to an increase of job satisfaction.
As we are now again able to do what we IT folks love – to build new solutions!
Besides on the really important fact of getting the IT back in-house we do also have a couple of other achivments through out our transition to AWS:
We can now deploy downtime free our software on a daily basis instead of having 3 major releases per year – which lead to a Reduced time to market
<CLICK>
The overall package of functionalities the cloud provides supports also an Agile culture: of exploration and of providing new platform functionalites faster and be prepared for any grows.
<CLICK>
We are now able to build solutions on state of the art technologies and are not limited any more to a given tech stack
<CLICK>
And last but not least: We are now able to adequately support new business opportunities like for example online sales where among others high availability is an import factor.
Summarize by encouraging people to download the session
Formed in January 2016
Investigated various providers, selected AWS as best fit
What really matters is autonomy for people. We’re working to improve lives by giving everyone the freedom to move – whether across town or across the room.
We are applying artificial intelligence to help Toyota produce cars in the future that are safer, more accessible and more environmentally friendly.
We are expanding artificial intelligence technology into new applications by strengthening and refining the interaction between human and machine
How safe is safe enough?
We believe automated vehicle technology will help save lives. TRI has been taking a dual-path approach to automated driving, Guardian and Chauffeur modes. Guardian and Chauffeur modes are being designed to give consumers a choice.
In the Guardian approach, the human driver maintains vehicle control, and the automated driving system operates in the background, monitoring for potential crash situations. It can intervene to help protect vehicle occupants when needed.
Chauffeur is TRI’s version of full vehicle autonomy where all occupants are passengers as the car drives itself.
Rich sensor suite – often have multiple, cameras, lidar, radar sensors
Generate 1 TB per hour / car of data
How much data is enough?
https://devblogs.nvidia.com/training-self-driving-vehicles-challenge-scale/
1 Tb/hr -> 2.2Gbit/s
Netflix Ultra HD stream is ~25Mbit/s (https://help.netflix.com/en/node/306)
How do you scale out to process all the data once you’ve collected it?
Meet the demands of all potential workloads?
How do you stay up to date with hardware advances?
- Technology continues to change, do you want to invest in hardware that has a 3 or 5 year depreciation, but might no longer meet your needs in 2 years?
Build: Experiment and build a promising model on small dataset
CI: Make sure code base is free of defects and bugs
Train and tune: Make the model work with larger datasets and optimize it
Prepare a model for deployment and validate it
Build: Experiment and build a promising model on small dataset
CI: Make sure code base is free of defects and bugs
Train and tune: Make the model work with larger datasets and optimize it
Prepare a model for deployment and validate it
Deploy model to vehicles
The lower your iteration times, the more productive your researchers and developers.
Talk about log processing pipeline
Talk about clusters
Pytorch
Lots of services to take advantage of, lets you focus on what makes your company different
Leverage others- AWS – truly are customer focused. Work with your account team to get new feature requests in, understand problems, help. ProServe
F7: We knew what we wanted to do with IaC and pipelines
F7 is excellent at this and accelerating our momentum while we work on the unique things to our company
Going back to the “how safe is safe enough” question, if we assume for now that an AV must be 10 times as safe as human driving, that would mean that the fatality rate of AV must be one in 1 trillion miles.
How do we get there and also prove that statistically?
Testing of our prototype vehicles on public roads is certainly not enough. We would need thousands of vehicles running 24 hours and it would still take years and years to prove the safety, because we can never cover every possible traffic scenario.
Therefore, we are using simulation to virtually re-create traffic situations in a computer and run our software, adding adverse conditions such as bad weather and difficult traffic pattern.
We are also putting a lot of effort into gathering data from real drivers so we can train our AI.
Sanat – we still have a lot of great automotive content remaining that I encourage you to attend.