Design a unique ID generator in HLD - System Design Guide

Problem Statement

When multiple systems or services generate IDs independently, collisions can occur, causing data corruption or overwrites. Without a reliable unique ID generator, distributed systems struggle to maintain data integrity and traceability across components.

Solution

A unique ID generator creates identifiers that are guaranteed to be distinct across time and space. It can use techniques like timestamp-based components, machine identifiers, and sequence numbers to ensure uniqueness even in distributed environments. This allows systems to generate IDs without coordination, avoiding bottlenecks and collisions.

Architecture

Client/Service

Instance 1

→Unique ID Gen

↓

Client/Service

Instance 2

→Unique ID Gen

This diagram shows multiple client or service instances requesting unique IDs from their local unique ID generator instances, which then store or use these IDs in a database. Each generator instance ensures IDs are unique without central coordination.

Trade-offs

✓ Pros

→

Enables distributed ID generation without a single point of failure or bottleneck.

→

Supports high throughput by allowing parallel ID generation across multiple nodes.

→

Ensures IDs are sortable by time if timestamp components are included.

✗ Cons

→

Requires careful design to avoid collisions, especially with clock skew or machine ID conflicts.

→

Complexity increases with the need to coordinate machine identifiers or handle clock rollback.

→

Debugging ID generation issues can be difficult in distributed environments.

Use when your system is distributed across multiple nodes or data centers and requires scalable, collision-free ID generation at high request rates (e.g., over thousands of IDs per second).

Avoid if your system is small-scale with a single database generating auto-increment IDs, or if strict sequential IDs are mandatory without gaps.

Real World Examples

Twitter

Twitter uses Snowflake, a distributed unique ID generator that encodes timestamp, machine ID, and sequence number to generate unique, sortable tweet IDs without coordination.

Instagram

Instagram uses a similar approach to generate unique IDs for posts and comments across distributed servers, ensuring no collisions and enabling efficient data sharding.

Amazon

Amazon uses unique ID generators to assign order IDs and transaction IDs across multiple services and regions, ensuring global uniqueness and traceability.

Alternatives

Centralized ID Generator

All ID requests go to a single central server that issues IDs sequentially.

Use when: Choose when system scale is low and simplicity is preferred over distributed scalability.

UUID (Universally Unique Identifier)

IDs are generated using random or pseudo-random numbers without coordination, relying on probability for uniqueness.

Use when: Choose when you need simple, decentralized ID generation without ordering guarantees.

Database Auto-Increment IDs

IDs are generated by the database as sequential numbers on insert.

Use when: Choose when a single database handles all writes and strict sequential IDs are required.

Summary

Unique ID generators prevent collisions in distributed systems by combining timestamps, machine IDs, and sequences.

They enable scalable, parallel ID creation without a central bottleneck or coordination.

Choosing the right ID generation pattern depends on system scale, ordering needs, and complexity tolerance.

Practice

1. What is the primary purpose of a unique ID generator in a distributed system?

easy

A. To create identifiers that are distinct across all machines and time

B. To encrypt data for secure communication

C. To compress large files efficiently

D. To balance load between servers

Solution

1. Step 1: Understand the role of unique IDs

   Unique IDs ensure that each identifier is different from others, avoiding conflicts.

2. Step 2: Recognize distributed system needs

   In distributed systems, IDs must be unique across machines and time to prevent collisions.

3. Final Answer:

   To create identifiers that are distinct across all machines and time -> Option A

4. Quick Check:

   Unique ID purpose = distinct identifiers [OK]

Hint: Unique IDs prevent duplicates across systems [OK]

Common Mistakes:

• Confusing unique ID with encryption
• Thinking unique ID compresses data
• Mixing load balancing with ID generation

2. Which of the following is a common component in a unique ID generator design?

easy

A. Encryption key for data security

B. Load balancer to distribute requests

C. Compression algorithm for data size reduction

D. Sequence number to avoid collisions within the same timestamp

Solution

1. Step 1: Identify components of unique ID generators

   Common components include timestamp, machine identifier, and sequence number.

2. Step 2: Understand sequence number role

   Sequence numbers help generate multiple unique IDs within the same timestamp to avoid collisions.

3. Final Answer:

   Sequence number to avoid collisions within the same timestamp -> Option D

4. Quick Check:

   Sequence number = collision avoidance [OK]

Hint: Sequence numbers prevent same-time ID clashes [OK]

Common Mistakes:

• Confusing encryption with ID generation
• Thinking compression is part of ID design
• Mixing load balancing with ID components

3. Consider a unique ID generator that uses a 41-bit timestamp, 10-bit machine ID, and 12-bit sequence number. What is the maximum number of unique IDs it can generate per millisecond per machine?

medium

A. 8192

B. 1024

C. 4096

D. 2048

Solution

1. Step 1: Understand bit allocation for sequence number

   The sequence number uses 12 bits, so max IDs per millisecond = 2^12.

2. Step 2: Calculate 2^12

   2^12 = 4096 unique IDs per millisecond per machine.

3. Final Answer:

   4096 -> Option C

4. Quick Check:

   2^12 = 4096 [OK]

Hint: 2^sequence_bits = max IDs/ms [OK]

Common Mistakes:

• Using machine ID bits instead of sequence bits
• Calculating 2^10 or 2^11 instead of 2^12
• Confusing total bits with sequence bits

4. A unique ID generator uses a timestamp, machine ID, and sequence number. If two machines generate IDs at the exact same millisecond with the same sequence number, what is the likely cause of duplicate IDs?

medium

A. Machine IDs are not unique or not included in the ID

B. Timestamp is too large

C. Sequence number is too long

D. The system uses encryption

Solution

1. Step 1: Analyze ID components for uniqueness

   Machine ID differentiates IDs from different machines at the same time.

2. Step 2: Identify cause of duplicates

   If machine IDs are missing or not unique, IDs from different machines can collide.

3. Final Answer:

   Machine IDs are not unique or not included in the ID -> Option A

4. Quick Check:

   Missing unique machine ID = duplicates [OK]

Hint: Unique machine ID prevents cross-machine duplicates [OK]

Common Mistakes:

• Blaming timestamp size for duplicates
• Thinking longer sequence number causes duplicates
• Confusing encryption with ID uniqueness

5. You need to design a unique ID generator for a global system with thousands of machines generating millions of IDs per second. Which design choice best ensures scalability and uniqueness?

hard

A. Generate random 64-bit numbers without coordination

B. Use a 64-bit ID combining timestamp, machine ID, and sequence number with synchronized clocks

C. Use only timestamp-based IDs without machine info

D. Assign IDs sequentially from a central server

Solution

1. Step 1: Consider scalability and uniqueness needs

   Global scale requires IDs unique across machines and time, with high throughput.

2. Step 2: Evaluate design options

   Combining timestamp, machine ID, and sequence number in 64 bits with synchronized clocks ensures uniqueness and scalability.

3. Step 3: Reject other options

   Random IDs risk collisions; timestamp-only lacks machine uniqueness; central server causes bottleneck.

4. Final Answer:

   Use a 64-bit ID combining timestamp, machine ID, and sequence number with synchronized clocks -> Option B

5. Quick Check:

   64-bit composite ID = scalable unique IDs [OK]

Hint: Combine time, machine, sequence for scalable unique IDs [OK]

Common Mistakes:

• Relying on random IDs risking collisions
• Ignoring machine ID causing duplicates
• Using central server causing bottlenecks