Explaining distributed systems like I'm five

Michele Riva

The idea for this talk started during an online course

The idea for this talk started during an online course

Gabriele

Me

Friends

What problems do distributed systems solve?

Create an API that uploads a large CSV file (>10GB) to a server, then read the file, normalize the data, and store it into a database

What if we have dozens of concurrent uploads?

Once we start uploading and processing a few large files, we'll run out of memory

The ice cream scenario

How do we currently deal with large amounts of people?

With queues!

With queues!

But what if there's a *lot* of people?

We hire a security guard

When it comes to software architecture, things gets a bit harder...

A collection of indipendent computers that appear to its users as one computer

A distributed system is

(Andrew Tanenbaum)

• Computers operate concurrently

• Computers operate concurrently
• Computers fail independently

• Computers operate concurrently
• Computers fail independently
• Computers do not share a global clock

Source of truth

Source of truth

Sync request

Sync request

Source of truth

Time response

Time response

Source of truth

Time response

Time response

Network time protocol (NTP)

Precision time protocol (PTP)

• Computers operate concurrently
• Computers fail independently
• Computers do not share a global clock

The solution needs to follow the following rules

Is uploading a large file to a server a good option?

• Do computers operate concurrently?
• Do computers fail independently?
• Do computers share a global clock?

Have we respected the rules?

So what problems did we solve?

Michele Riva