LLD: Design S3

If you’re looking for 1:1 mentorship with a strong focus on LLD (core emphasis on Multithreading), HLD, DSA, and system design research papers, feel free to reach out.

📩 Contact: programmingappliedai@gmail.com

FR

• Create buckets

• Put/Get/Delete objects

• Object versioning

• Thread-safe (basic)

• Simple APIs similar to S3

🧱 Core Entities

S3Service
 ├── Bucket
 │     ├── ObjectKey
 │     │      └── List<ObjectVersion>

---

🧩 Java Code (Executable LLD)

1️⃣ ObjectVersion

import java.time.Instant;

class ObjectVersion {
    private final String versionId;
    private final byte[] data;
    private final Instant createdAt;

    public ObjectVersion(String versionId, byte[] data) {
        this.versionId = versionId;
        this.data = data;
        this.createdAt = Instant.now();
    }

    public String getVersionId() {
        return versionId;
    }

    public byte[] getData() {
        return data;
    }
}

---

2️⃣ S3Object

import java.util.*;

class S3Object {
    private final String key;
    private final List<ObjectVersion> versions = new ArrayList<>();

    public S3Object(String key) {
        this.key = key;
    }

    public synchronized void addVersion(ObjectVersion version) {
        versions.add(version);
    }

    public synchronized ObjectVersion getLatestVersion() {
        if (versions.isEmpty()) return null;
        return versions.get(versions.size() - 1);
    }

    public synchronized ObjectVersion getVersion(String versionId) {
        return versions.stream()
                .filter(v -> v.getVersionId().equals(versionId))
                .findFirst()
                .orElse(null);
    }
}

---

3️⃣ Bucket

import java.util.concurrent.ConcurrentHashMap;

class Bucket {
    private final String name;
    private final ConcurrentHashMap<String, S3Object> objects = new ConcurrentHashMap<>();

    public Bucket(String name) {
        this.name = name;
    }

    public void putObject(String key, byte[] data) {
        objects.putIfAbsent(key, new S3Object(key));
        String versionId = UUID.randomUUID().toString();
        objects.get(key).addVersion(new ObjectVersion(versionId, data));
    }

    public byte[] getObject(String key) {
        S3Object obj = objects.get(key);
        if (obj == null) return null;

        ObjectVersion version = obj.getLatestVersion();
        return version != null ? version.getData() : null;
    }

    public void deleteObject(String key) {
        objects.remove(key);
    }
}

---

4️⃣ S3Service (Main API)

import java.util.concurrent.ConcurrentHashMap;

class S3Service {
    private final ConcurrentHashMap<String, Bucket> buckets = new ConcurrentHashMap<>();

    public void createBucket(String bucketName) {
        buckets.putIfAbsent(bucketName, new Bucket(bucketName));
    }

    public void putObject(String bucketName, String key, byte[] data) {
        Bucket bucket = buckets.get(bucketName);
        if (bucket == null) {
            throw new RuntimeException("Bucket not found");
        }
        bucket.putObject(key, data);
    }

    public byte[] getObject(String bucketName, String key) {
        Bucket bucket = buckets.get(bucketName);
        if (bucket == null) return null;
        return bucket.getObject(key);
    }

    public void deleteObject(String bucketName, String key) {
        Bucket bucket = buckets.get(bucketName);
        if (bucket != null) {
            bucket.deleteObject(key);
        }
    }
}

---

5️⃣ Driver Code (Test)

public class Main {
    public static void main(String[] args) {
        S3Service s3 = new S3Service();

        s3.createBucket("photos");

        s3.putObject("photos", "img1.jpg", "image-data-1".getBytes());
        s3.putObject("photos", "img1.jpg", "image-data-2".getBytes());

        byte[] data = s3.getObject("photos", "img1.jpg");
        System.out.println(new String(data));

        s3.deleteObject("photos", "img1.jpg");
    }
}

---

🔍 Key LLD Concepts Used

• Encapsulation → Bucket owns objects

• Concurrency → ConcurrentHashMap

• Versioning → List of ObjectVersion

• Single Responsibility

• API-first thinking (like real S3)

---

🧠 Interview Extensions (Say this out loud)

If interviewer asks:

• How to scale? → Partition buckets, metadata DB + object storage (SSTables)

• Large files? → Multipart upload

• Durability? → Replication + checksum

• Auth? → IAM + signed URLs

• Consistency? → Read-after-write for new objects

---