Requirements of the Typeahead Suggestion System’s Design

Scope the suggestion behavior before sizing it. The number of completed searches differs from the number of prefix requests, and a response contains several complete suggestions rather than one character.

Requirements

We will design a typeahead suggestion system that meets the following functional and non-functional requirements.

Functional requirements

The system suggests the top $N$ frequent and relevant terms based on the user’s input.

Non-functional requirements

• Low latency: Use 200 ms as an initial end-to-end latency target for this exercise, then choose a percentile and test it under load. Typing speed varies; 160 ms is an illustrative interval, not a universal average. Tag requests so a late response for an older prefix cannot replace current suggestions.

• Fault tolerance: The system must provide suggestions even if individual components fail.

• Scalability: The system must support an increasing number of users and queries over time.

Resource estimation

Use a hypothetical workload of 3.5 billion completed searches per day. This is a sizing assumption, not a current Google statistic. Assume 15 characters per query and initially one suggestion request per typed character; later client optimizations can reduce that request count.

Storage estimation