Daimler will present the Mercedes Telediagnosis use case where MongoDB was chosen as the storage document database.
You will learn some basics about Telediagnosis, the requirements which drove us to MongoDB, the advantages we achieved, and our experiences on our MongoDB journey.
2. Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
2
About Myself
Name: Mădălin Broscaru
Role: IT Architect - Diagnostics and Connected Car Data/ Daimler Aftersales
Topics: Vehicle Diagnosis / Telediagnosis
Aftersales Connected Services
Connected Cars
3. 3
Agenda
• Present our MongoDB use case: Telediagnosis @Daimler
• Why Daimler has chosen MongoDB
• Data specifics
• How data is accessed
• Data storage requirements
• Our Mongo DB journey
• How we started with MongoDB (sizing)
• How our Architecture looks like (Sharding, Replicas, Indexes)
Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
4. Seamless integrated mobility platforms will be game changer
for future mobility services
Connected Vehicles: Digital Transformation requires Game Changers in IT-Support and Operations of Mercedes Benz Cars | Ruediger
Schmid, Daimler AG
4
Mercedes me connect
E-Mobility
Parking Services
Car 2 X Communication
Autonomous Driving
Car Sharing
Smart Home
User Device Connectivity
Mobility Services
me
CASE
5. 5
Telematics as Core Process in the future Mercedes Services
Transforming data into business opportunities (e.g. Mercedes connect services, data driven business
models, autonomous driving)
… the Vehicle Data
Conditioning (VDC)
where these technical
vehicle events are
processed … CAC
Retail
Customer
… and the follow-up
processes are triggered with
real-time recommendations for
actions.
VDC
Vehicles are transmitting regularly
status and health data into ...
Agregated
Quality Analisys
Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
8. 8
How Data is accessed
Controlling Units
Failures
WDD176039 j540918A84
N10-Signal acquisition module
(SAM)
MPC-Multifunction Camera
Water fluid level too low
Today 22 October 2019 9:40
WDD176039 j540918A84
Today 22 October 2019 9:40
MPC-Multifunction Camera
Error Frequency: Counter: 3
Ignition Cycle: Counter: 9
Software Version: 15/0930
Hardware Revision:13/0930
Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
9. 9
Top Requirements
• Horizontal Scaling
• Support HA
• Speed / Fast Response Time
• Mature technology, good community
• Cost
Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
10. 10
Our Journey with MongoDB: First POC - 1
MongoDB –
primary
• Ubuntu
• 4 Cores
• 32 GB
• 500 GB SSD 3000
IOPS
MongoDB –
secondary
• Ubuntu
• 4 Cores
• 32 GB
• 500 GB SSD 3000 IOPS
JSON
Producer
MongoDB –
secondary
• Ubuntu
• 4 Cores
• 32 GB
• 500 GB SSD 3000 IOPS
Data
Reader
• How easy it is ?
• Storage requirements for
500 mil EES
• Test MongoDB
performance
• Insight on scaling
Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
15. 15
• Sharding
• Using 3 shards from the begining.
• Expecting to store up to 2 TB per shard
• 3 Replica set node per shard
• Using the VIN (chassisNumber) as the shard key
• Indexing
• VIN (chassisNumber), ECU, DTC
Mongo DB - Project Specifics
• Collections
• Data Migration
• Backup
• Cloud vs On-Premise
• Monitoring
• Operations
Telediagnosis@Daimler powered by MongodDB| Madalin Broscaru,
Daimler AG
17. Seamless integrated mobility platforms will be game changer
for future mobility services
Connected Vehicles: Digital Transformation requires Game Changers in IT-Support and Operations of Mercedes Benz Cars | Ruediger
Schmid, Daimler AG
17
@Daimler
Glad to be here with you today
Few words about myself
A quick overview of what I‘m going to show you today
Before solution, what is specific to requirement is relevant
Few words about context
Auto industry
Middle transformation
From Future Mobility one lever deeper
The telediagnosis usecase where MongoDB comes
What happens beyound the scene
You saw the domain, but how the data looks like?
What kind of data a car has
Car as a mini datacenter up to > 100 ECUs
Engine Control Module, Brake Control Module, Suspension Control Module, Clima, Inteligent Lighting, Clima, Adaptative Driv
This is how data is looking, oversimplified, human readable
Diagnostic Trouble Code
Saw the data but what about the access?
Relevant for sharding, indexes
+ Cust. Assistance Center
+ Mercedes Me
+ Automated valet parking
Beside all data and access we have also NFRS
NFRs ware the triggers for MongoDB, Data and access are enablers
Do the test drive before you buy
Parenting books from people without children
Why not as a service
Transition:
And base on the POC we got our own insights.
Inserted 62 mil (20 h)
W -7-10. R-3-6
Indexes best practices: in memory, serch by index
compression 50% W I R E D Tiger
Not so many deletes and updates
Cloud watch, see Bootleneck: CPU, RAM, NETWORK, IO
My advice, make your own experience with it, it helps you understand better the product
We like what we saw with POC so we move MVP
Some things are reusable.
Idea:
Idea what is below is the productive env
What is above is the operations
Transition:
What happens if we have multiple env?
Idea:
in our case, test = prod, only smaller. Why?
You saw a little on the infrastructure and deployment archtecture
What about Coniguration Specifics?
This is where we are and what worked for us
We are at the beginning (no so much aggregation, relations)
It will come more and more requirement
Mongo DB is used to implement the future mobility @ Daimler