Design top-K frequency store for varying workloads

Scenario

You need to design an in-memory component that tracks how often each key appears in a stream of events.

The component must support these operations:

• record(key) : increment the count for key by 1 (insert key if not present).
• topK(k) : return the k keys with the highest counts so far (order among them does not matter).

Assume:

• There can be up to N distinct keys.
• Operations happen online as the stream arrives.

Task

Describe how you would implement this component under two different workload patterns:

1. Read-heavy : there are many more topK(k) calls than record(key) calls.
2. Write-heavy : there are many more record(key) calls than topK(k) calls.

For each case, specify:

• The main data structures you would use.
• Time complexity of record and topK .
• Space complexity.
• Why your design is appropriate for that workload (the trade-offs you’re making).