System requirements

Functional:

-> Vehical entry and exit management

-> Parking space optimisation

-> User Authentication and Payment

-> Real-Time Parking Availability

->Security Monitoring

->ERT

-> Admin dashboard

Non-Functional:

1. Total number of parking slots to be maintained
2. small car: 50%
3. Large car : 30%
4. Morotcycle: 10%
5. Special abled: 5%

• Small Cars: 50% of 500 = 250 slots
• Large Cars: 30% of 500 = 150 slots
• Motorcycles: 10% of 500 = 50 slots
• Handicap Accessible Spaces: 5% of 500 = 25 slots

2.Average influx and outflux

Avg influx : 1 vehical/minute

Avg outflux : 1.5 vehical/minute

3.Slot stauts update event rate 1 event/second on a slot

4. System should be highly available

5.System should be resilient to failure

6.Secure

7.Performent

8.Parking slot provided should be highly consistent.

1. Slot status update events 1 event/second

500 * 1 event/sec= 500 event/second

1. Ticket generation API 10 id/minute
2. Payment api 10 id/minute
3. 100 days of history needed . Lets assume in a parking slot on a given day 24 vehicals get parked

24 * 500 * 100 = History events

Capacity estimation

Storage for slots

1 slot need 1Mb

500 * 1Mb = 500 Mb storage needed for slots

Occupancy event storage

1 event/second per parking slot

500 events/seconds

State change event storage

24 event/day * 500 = 12000 /day

1 month state change history need = 30 * 12000= 360000 events

1 state change event = 10 kb

total storage for 1 month= 36Gb

Stage change event processing speed

We need a real time update of the parking slot we should process 500 events/seconds within 1 second and final state should be updated in resulting storage.

API design

-> API to onboard vehicle

HTTP POST

v1/parking/onboard

Request: {

vehicalNumber: "",

VahicalImage: []byte,

EntryTime:DateTime,

}

Response: {

id:UUID,

entryTime:""

VehicalNumber:""

ParkingSlot:""

NavigationPath:[]

}

-> API to exit from parking slot

HTTP POST

v1/parking/exit

{

id:UUID,

entryTime:""

exitTime:"'

VehicalNumber:""

ParkingSlot:""

NavigationPath:[],

paymentMode:{

txnId:"",

amount:""

}

}

response:

{

id:UUID,

entryTime:"",

exitTime:"',

VehicalNumber:"",

ParkingSlot:"",

exitMessage:""

}

Database design

-> Storing parking slot information. Parking slot show transactional behaviour so i will chose RBDMS for storing parking slot data.

-> Storing parking slot event change: Events once stored will not be modified but it keeps appending. But may need some analytical operation on top of it. I need a Horizontally scalable DB(Sharded ) and Higlhy available(replicated) and which allows search, filtering and sorting operation on the content.

MongoDB is one of the good option.

-> We need a storage for payment history done by user.

High-level design

You should identify enough components that are needed to solve the actual problem from end to end. Also remember to draw a block diagram using the diagramming tool to augment your design...

Request flows

CheckIN flow

1. API call reach Load balancer where TLS termination happens and later call will be forwarded to API-handler.
2. API-handler does request validation and forward the call to slot-manager service.
3. Slot-manger does the optimisation job and allocate appropriate slot to the vehicle. Update the parking slot DB and mark slot as occupied.
4. Slot manager generate the ID for vehicle and respond to API-handler.

Optimal Parking slot calculator flow. This is to derive the optimal number of parking slots in various space of the system

1. Through API's admin will provide input of rectangular space of patched available in the parking lot.
2. call will reach API-handler then its moved to ParkingSlotOptimizer
3. ParkingSlotOptimizer will use optimization algorithm to find optimum number of car, truck, Bike parking that can be created in the given rectangle patch.

Event Flow:

1. Noticiation service receives the telemetry
2. Telemetry are processed asynchronously . Telemetry will be queued.
3. Notification processor wait for flip in the event and event change will be presisted in the event database. If event is same it will be dropped.
4. Event change can be stored in Time series database for more insightful analytics later.

Detailed component design

ParkingSlotOptimizer

1. Give a rectangle area , Number of small, medium and heavy vehicle expected in the system ParkingSlotOptimizer will come up with optimal parking slot allocation. Its a typical knapSack kind of optimisation problem.

Notification Service

Internally it receives events every 1 seconds based on the state provided by parking lot sensors. Noticiation service store last event in in-memory cache like redis (sharded and replicated) on every new event it will check event stored in DB and if there is delta/change then new event is persisted in the event-state-change DB.

We need efficient diff alogorithm to find event diff.

Trade offs/Tech choices

For parking slot store RDBMS is chosen since data is transactional.

Failure scenarios/bottlenecks

RDBMS will have replicated read-replicas but if primary crash there will be down time during auto recovery

Future improvements

What are some future improvements you would make? How would you mitigate the failure scenario(s) you described above?