Why Your System Breaks at 100 Users (Not 1)

Everything works perfectly.

With 1 user.

Then you get 100 users…

And suddenly:

• APIs slow down
• Database spikes
• Errors appear

Nothing changed in your code.

So what broke?

👉 Your assumptions.

---

The Illusion

Most developers test like this:

• Local machine
• Single request
• Small dataset

Everything looks fast.

But production is different:

• Many users
• Concurrent requests
• Shared resources

---

Problem 1: N+1 Queries (Silent Killer)

What You Write

const blogs = await db.select().from(blogs);

for (const blog of blogs) {
  const author = await db
    .select()
    .from(users)
    .where(eq(users.id, blog.authorId));
}

What Happens

• 1 query to fetch blogs
• N queries for authors

👉 Total = N + 1 queries

If N = 100 → 101 queries per request

---

Fix: Use JOIN

const result = await db
  .select({
    title: blogs.title,
    author: users.name,
  })
  .from(blogs)
  .leftJoin(users, eq(blogs.authorId, users.id));

👉 1 query instead of 101

---

Problem 2: Database Connection Limits

Each request uses a DB connection.

Example:

• 100 users
• Each request = 1 connection

👉 You hit DB limits quickly

---

What Happens

• New requests wait
• Queries slow down
• System appears "down"

---

Fix: Connection Pooling

Use pooled connections instead of opening new ones.

const client = postgres(connectionString, {
  max: 10,
});

👉 Reuse connections instead of creating new ones

---

Problem 3: Blocking Operations

Bad pattern:

await createOrder();
await sendEmail();
await updateAnalytics();

👉 Everything blocks the request

---

Fix: Background Jobs

await createOrder();

await queue.add("send-email", { userId });
await queue.add("analytics", { orderId });

👉 Fast response + async processing

---

Problem 4: Over-fetching Data

Bad:

const blogs = await db.select().from(blogs);

👉 Fetches everything

---

Fix: Select Only Needed Fields

const blogs = await db
  .select({
    title: blogs.title,
    slug: blogs.slug,
  })
  .from(blogs);

👉 Smaller payload → faster response

---

Problem 5: No Indexing

Without index:

SELECT * FROM blogs WHERE slug = 'post';

👉 Full table scan

---

Fix: Add Index

CREATE INDEX idx_slug ON blogs(slug);

👉 Fast lookup

---

Summary Table

Problem	Impact	Fix
N+1 Queries	Slow DB	JOIN
Connection Limits	Timeouts	Pooling
Blocking Tasks	Slow APIs	Queues
Over-fetching	Large payloads	Select fields
No Indexing	Slow queries	Index

---

Key Insight

Your system doesn't break because of traffic.

It breaks because of inefficiencies multiplied by traffic.

---

Final Takeaway

1 user hides problems.

100 users expose them.

---

Closing Thought

Before scaling servers, ask:

👉 "What happens when 100 requests hit this at the same time?"

If the answer is unclear…

👉 your system will break.