A practitioner's guide to building scalable systems

Building Scalable Systems: Distributed Systems basics to System Design Interviews

We want to give you the tools to compare multiple approaches and pick the one that works for your scenario.

Grokking the Distributed and Scalable Systems

# Layers in the OSI Model

_This might be redundant for you if you remember it from your Networking 101_

Open Systems Interconnection (OSI) is a conceptual framework that engineers use to model communication between computer systems. This logical model describes how computers can talk to each other. It has seven layers.

1. Application
2. Presentation
3. Session
4. Transport
5. Network
6. Data Link
7. Physical

It is typically compared with the TCP/IP model which is how we usually think about transmission between two computers.







Out of the seven layers, we are primarily interested in the following two:

1. Application Layer
2. Transport Layer

## What is the Application Layer
This layer includes applications with which the user interacts with e.g. your browser, Skype, Outlook, etc.

Typical protocols that operate at this layer are HTTP, SMTP, FTP, DNS, etc.

## What is the Transport Layer?
Transport layer deals with splitting the data into multiple packets to send it across the network. It also takes care of reconstructing the data at the destination for the layers above it.

This layer also deals with ensuring the delivery of data while maintaining the integrity of data. 

Typical protocols at this layer are TCP, UDP, and many others.

## What does it have to do with Load Balancers

Good news is that you don't have to dig deeper into what each layer does. Just remember two things.

1. At Layer 4, we don't have access to the application data e.g. session_id/cookies, etc.

2. At Layer 7, we have access to application data.

_With that, let's look at L4 and L7 load balancers._

# Layer 4 (L4) Load Balancers

Layer 4 Load Balancers make their routing decisions based on the information provided by the first few bytes of the TCP e.g. Source IP address and Port. They don't look at the application data.

After looking at the TCP/IP information of the request, it makes a routing decision and then uses Network Address Translation (NAT) to forward the packet to a particular server.

Traditionally, L4 Load Balancers were more popular as they require less computing resources and are quite efficient at being able to forward thousands of requests per second.

# Layer 7 (L7) Load Balancers

Layer 7 Load Balancers make their routing decisions by looking at the application data e.g. things like 

1. The URL of the request
2. Http Header
3. The Type of the data (text/image)
4. Information stored in the cookie.
5. Session ID

L7 Load Balancers require more computing power. However, as raw processing power is becoming cheaper, L7 routers are becoming more popular.

L7 load balancers can make smarter decisions e.g. they can recognize that a certain request is for an image that's already available in a cache and forward the request to that caching server etc.


_If you were paying close attention, we've said multiple times that the Load Balancer "decides" to send traffic to a particular server. How does it decide? That's what we are going to tackle next._

What and Why

Node in a Distributed System

Performance and Numbers

Fault Tolerance and Availability

Understanding TIME in Distributed Systems

Data Centers for software engineers

Scaling Distributed Systems

Load Balancers

Structured Storage at Scale

Sharding in Storage Systems

SQL vs NO-SQL

Blob Storage at Scale

Caching in Distributed Systems

Queues and Asynchronous Processing

Data Analytics and Batch Processing

Logging and Monitoring

Searching at Scale

Conclusion

Layer 4 v.s Layer 7 Load Balancers

Layers in the OSI Model