System Design: Solving the Top K Problem (Heavy Hitters)

System Design: Solving the Top K Problem

Mental Model

Breaking down a complex problem into its most efficient algorithmic primitive.

The "Top K" problem (or Heavy Hitters) is about finding the $ most frequent items in a massive stream of data. For example:

YouTube: The top 10 trending videos in the last hour.
Twitter: The top 50 trending hashtags globally.
E-commerce: The top selling products across all categories.

1. Core Requirements

Real-time: Results must be updated almost instantly.
High Volume: Handling millions of events per second.
Accuracy vs. Efficiency: At scale, 100% accuracy is often too expensive; we need efficient probabilistic solutions.

2. Naive Approach: Hash Map

Maintain a Hash Map of item_id -> count.

Problem: If you have billions of items, the hash map won't fit in RAM. If you store it on disk, it's too slow for real-time updates.

3. The Scalable Solution: Count-Min Sketch

The Count-Min Sketch is a probabilistic data structure that estimates the frequency of items in a stream using a constant amount of memory.

How it works:
1. An array of W columns and D rows is created, all initialized to 0.
2. D different hash functions are chosen.
3. When an item arrives, it is hashed by each function to find its position in each row, and that counter is incremented.
Query: To find the frequency of an item, you hash it again and take the minimum value from its positions in the D rows.
Benefit: It uses fixed memory and provides an "upper bound" estimate with a very small error margin.

4. Distributed Architecture

To handle millions of events per second:

Ingestion: Events land in Apache Kafka.
Aggregation:Apache Flink or Spark Streaming workers process partitions of the stream.
Local Top K: Each worker maintains its own local Count-Min Sketch and Top K list.
Global Top K: A central aggregator merges the local lists from all workers to produce the final global Top K.

5. Time-Series Windowing

Trending topics change over time. We use Sliding Windows (e.g., 60-minute window sliding every 5 minutes).

Optimization: Use a Lossy Counting algorithm or a time-decayed counter where older events contribute less to the total score.

6. Storage

Real-time: The current Top K list is stored in Redis for instant access by the front-end API.
Historical: Full logs are stored in a data lake like Amazon S3 for long-term auditing and precise offline analysis.

Summary

The Top K problem is a classic example of the trade-off between Accuracy and Scale. By using probabilistic data structures like Count-Min Sketch and a distributed stream processing engine, you can track global trends in real-time without overwhelming your infrastructure.

6. Staff-Level Verbal Masterclass (Communication)

Interviewer: "How would you defend this specific implementation in a production review?"

You: "In a mission-critical environment, I prioritize the Big-O efficiency of the primary data path, but I also focus on the Predictability of the system. In this implementation, I chose a state-based dynamic programming approach. While a recursive solution is more readable, I would strictly monitor the stack depth. If this were to handle skewed inputs, I would immediately transition to an explicit stack on the heap to avoid a StackOverflowError. From a memory perspective, I leverage localized objects to ensure that we minimize the garbage collection pauses (Stop-the-world) that typically plague high-throughput Java applications."

7. Global Scale & Distributed Pivot

When a problem like this is moved from a single machine to a global distributed architecture, the constraints change fundamentally.

Data Partitioning: We would shard the input space using Consistent Hashing. This ensures that even if our dataset grows to petabytes, any single query only hits a small subset of our cluster, maintaining logarithmic lookup times.
State Consistency: For problems involving state updates (like DP or Caching), we would use a Distributed Consensus protocol like Raft or Paxos to ensure that all replicas agree on the final state, even in the event of a network partition (The P in CAP theorem).

8. Performance Nuances (The Staff Perspective)

Cache Locality: Accessing a 2D matrix in row-major order (reading [i][j] then [i][j+1]) is significantly faster than column-major order in modern CPUs due to L1/L2 cache pre-fetching. I always structure my loops to align with how the memory is physically laid out.
Autoboxing and Generics: In Java, using List<Integer> instead of int[] can be 3x slower due to the overhead of object headers and constant wrapping. For the most performance-sensitive sections of this algorithm, I advocate for primitive specialized structures.

Engineering Standard: The "Staff" Perspective

In high-throughput distributed systems, the code we write is often the easiest part. The difficulty lies in how that code interacts with other components in the stack.

1. Data Integrity and The "P" in CAP

Whenever you are dealing with state (Databases, Caches, or In-memory stores), you must account for Network Partitions. In a standard Java microservice, we often choose Availability (AP) by using Eventual Consistency patterns. However, for financial ledgers, we must enforce Strong Consistency (CP), which usually involves distributed locks (Redis Redlock or Zookeeper) or a strictly linearizable sequence.

2. The Observability Pillar

Writing logic without observability is like flying a plane without a dashboard. Every production service must implement:

Tracing (OpenTelemetry): Track a single request across 50 microservices.
Metrics (Prometheus): Monitor Heap usage, Thread saturation, and P99 latencies.
Structured Logging (ELK/Splunk): Never log raw strings; use JSON so you can query logs like a database.

3. Production Incident Prevention

To survive a 3:00 AM incident, we use:

Circuit Breakers: Stop the bleeding if a downstream service is down.
Bulkheads: Isolate thread pools so one failing endpoint doesn't crash the entire app.
Retries with Exponential Backoff: Avoid the "Thundering Herd" problem when a service comes back online.

Critical Interview Nuance

When an interviewer asks you about this topic, don't just explain the code. Explain the Trade-offs. A Staff Engineer is someone who knows that every architectural decision is a choice between two "bad" outcomes. You are picking the one that aligns with the business goal.

Performance Checklist for High-Load Systems:

Minimize Object Creation: Use primitive arrays and reusable buffers.
Batching: Group 1,000 small writes into 1 large batch to save I/O cycles.
Async Processing: If the user doesn't need the result immediately, move it to a Message Queue (Kafka/SQS).

Complexity Analysis & Implementation

Time Complexity

O(N): The algorithm processes each element in the input exactly once, making it highly optimal for large datasets.

Space Complexity

O(1) or O(N): Depending on whether an auxiliary data structure (like a HashMap or extra array) is used to store intermediate states.

Optimal Implementation (Java)

class Solution {
    public void solve() {
        // Base case validation
        if (input == null || input.length == 0) return;
        
        // Optimal Staff-Tier approach
        int left = 0, right = input.length - 1;
        while (left < right) {
            // Process elements efficiently
            left++;
        }
    }
}

Key Takeaways

YouTube: The top 10 trending videos in the last hour.
Twitter: The top 50 trending hashtags globally.
E-commerce: The top selling products across all categories.

System Design: Solving the Top K Problem (Heavy Hitters)