100,000 Notifications/sec
Sustained peak request volume during high-traffic events.
High-ThroughputMid Level
High-Throughput Push & SMS Notification Engine
Deliver billions of multi-channel transactional notifications with priority queues and rate limiting.
Target Scale
Engineering Scale & Performance SLAs
Target production parameters expected in a senior or staff interview round.
100 Million Active Devices
Daily active users generating read and write operations.
100 GB / day logs
Projected data ingestion and replication storage capacity.
High-priority alert < 2s
Strict end-to-end percentile latency SLA constraint.
Stage 01
Functional & Non-Functional Requirements
Establish clear problem boundaries before proposing architectural components.
Core System Capabilities
- Send notifications across Push (APNs/FCM), SMS (Twilio), and Email (SendGrid/Resend).
- Support priority classification (Critical OTP > Transactional > Promotional).
- User opt-out preferences and frequency capping.
Reliability & Latency SLAs
- Guaranteed at-least-once delivery for critical alerts.
- Strict rate-limiting and deduplication to avoid spamming users.
- Resilient fallback between downstream carrier providers.
Stage 02
Capacity Estimation Math
Step-by-step arithmetic conversions for QPS, storage, and bandwidth.
| Dimension | Calculation Formula | Estimated Result |
|---|---|---|
| Notification Throughput | 100M users × 5 notifications/day = 500M notifications/day | ~5,800 avg/sec (~100,000 Peak/sec during flash events) |
Stage 03
Multi-Tier Architecture & Component Topology
How requests navigate ingress gateways, application logic, caching, and persistence.
API Gateway · Redis Deduplication Cache
Validates payload, checks idempotency key, and rejects duplicate notifications within 10 minutes.
RabbitMQ / Kafka Priority Queues · Worker Pools
Separates Critical (OTP), Transactional, and Marketing queues into dedicated processing pools.
APNs/FCM Worker · Twilio / SMS Gateway
Manages carrier rate limits, token rotation, and third-party error retries.
Stage 04
Database Schemas & Partitioning Strategy
Entity models, indexing, and primary key partitioning.
PK: user_id
- user_id (UUID)
- channel_email (BOOLEAN)
- channel_push (BOOLEAN)
- frequency_cap (INT)
Cached in memory for instant validation.
Stage 05
Critical Architectural Trade-Offs
How to defend engineering compromises when challenged by interviewers.
Single Queue vs Dedicated Priority Queues
Option A: Multi-Tier Priority Queues (OTP, Transactional, Marketing)
Option B: Single Unified Kafka Topic with Priority Headers
Rationale: Dedicated queues prevent massive marketing campaigns from starving critical two-factor authentication SMS messages.
Technical FAQ
Frequently Asked Questions: High-Throughput Push & SMS Notification Engine
Key interview questions and conceptual defenses.
How do you prevent duplicate notifications on network retries?
Use client-provided idempotency keys cached in Redis with a 10-minute TTL to drop redundant triggers.
Simulate this architecture
Download ClawPad