Design a Distributed File System

1. Problem Statement

2. Architecture Discussion Map

Use this as one discussion aid, not a single correct answer. Your design should follow from the requirements, scale, and trade-offs you establish.

graph LR
    A["Design a Distributed File System"]
    A --> F1["Metadata master and client metadata caching"]
    A --> F2["Chunk size, leases, and the write pipeline"]
    A --> F3["Append semantics and the consistency model"]
    A --> F4["Rack-aware placement, heartbeats, re-replication, and rebalancing"]
    A --> F5["Master durability, failover, and the small-files problem"]

3. Key Focus Areas

• 1
  Metadata master and client metadata caching
• 2
  Chunk size, leases, and the write pipeline
• 3
  Append semantics and the consistency model
• 4
  Rack-aware placement, heartbeats, re-replication, and rebalancing
• 5
  Master durability, failover, and the small-files problem

4. What the Interviewer Looks For

The interviewer probes these areas, and the report grades the interview against them.

• Requirements and capacity model
• Metadata master and client caching
• Chunks, leases, and the write pipeline
• Append semantics and consistency
• Placement, failure detection, and re-replication
• Master durability, failover, and metadata scale

5. What Strong Candidates Should Demonstrate

• Decide what the metadata service owns and how clients reach data without it becoming a bottleneck.
• Order and replicate concurrent writes and appends, and state what readers see after a partial failure.
• Place, re-replicate, and rebalance replicas across racks as disks and servers fail.
• Make master state durable and recoverable, and know when one master stops scaling.