← Atlassian Interview InsightsSoftware Engineer

Atlassian·Software Engineer·Onsite - System Design / Architecture·Senior

Senior
Jun 2026

Summary

Atlassian system design round focused on load balancing for a collaborative document service. The question looked straightforward at first but kept unfolding into more edge cases than I expected.

Questions Asked (2)

Q1

You have a document-sharing platform where users can collaboratively edit and view docs. Requests go through a load balancer using round-robin. What breaks with this setup for this kind of workload?

System DesignTechnical Trade-offs
Author's notes

I started with the obvious stuff like cache locality and uneven load, but the interviewer kept pushing and I realized I'd glossed over the WebSocket angle entirely.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the workload characteristics: collaborative editing requires real-time synchronization, session affinity, and consistency. Then explain how round-robin load balancing breaks these requirements, focusing on state, consistency, and user experience. Conclude by suggesting appropriate alternatives like sticky sessions, consistent hashing, or a stateful architecture.

Pro tip: Mention that while sticky sessions can help, they are not sufficient for real-time collaboration; you also need a pub/sub or WebSocket layer to propagate changes. This shows you understand the full stack of collaborative editing.

1. Identify workload characteristics

Explain that collaborative editing involves long-lived connections, real-time updates, and shared state. Users expect to see each other's changes instantly and consistently.

2. Explain round-robin limitations

Round-robin distributes requests evenly but ignores session state. For collaborative editing, this means users may hit different servers, leading to inconsistent views and lost updates.

3. Discuss specific breakages

Detail problems: WebSocket connections may be dropped or re-routed, causing missed updates; concurrent edits may conflict without a central coordination point; and presence information (who's online) becomes inconsistent.

4. Propose solutions

Suggest alternatives: sticky sessions for session affinity, consistent hashing to route users by document ID, or a centralized collaboration service (e.g., using operational transformation or CRDTs) with a pub/sub backbone.

5. Summarize trade-offs

Acknowledge that while round-robin is simple and good for stateless services, collaborative editing requires stateful routing and synchronization. Highlight the need for a more sophisticated load balancing strategy.

Key Points to Mention

  • Session affinity (sticky sessions) to keep users on the same server
  • WebSocket and long-lived connections need consistent routing
  • Consistency models: eventual vs strong consistency for collaborative edits
  • Operational transformation (OT) or CRDTs for conflict resolution
  • Presence and awareness (who's editing) require shared state
  • Scalability challenges with stateful servers and potential single points of failure

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

Walk through how you'd fix each of those load balancing problems specifically for a collaborative editing service.

System DesignTechnical Trade-offsAPI & Integrations
Author's notes

This is where I felt more comfortable.

Create a free account to read the full note