This video explains connection management, a key part of backend system design. It talks about how clients connect to backend applications using things like TCP or UDP. The video also shows how backend applications often need to connect to other backend applications, like an API gateway connecting to another service or a server connecting to a database, highlighting the two types of socket connections: front-end and upstream.

Key Takeaways

Connection management is a fundamental concept in backend application system design.

Clients connect to backend applications primarily through a connection, often TCP or more recently QUIC over UDP.

Each client connection creates a socket connection object on the backend, which buffers and processes incoming requests based on various protocols like HTTP/2 or PostgreSQL.

Backend applications frequently connect to other backend applications, such as an API gateway forwarding requests to an upstream service, requiring a second, upstream socket connection.

A backend server connecting to a database server is another common example of an upstream socket connection, often using a different protocol.

0:00 Perhaps one of the most important system

0:03 designs

0:04 uh concept especially in backend

0:07 applications is uh connection

0:09 management.

0:11 Uh you see this object that is the

0:14 socket connection

0:17 lives on both the front end side of the

0:20 backend application and also at the

0:24 upstream side which is the back end side

0:26 of that. So what does that mean? uh you

0:29 see when a client want to connect to a

0:31 backend uh the only medium is through a

0:36 connection often a TCP connection most

0:39 recently it's a quick connection through

0:41 UDP

0:42 but then

0:45 that

0:47 front end connection that creates a

0:48 connection object on the back end for

0:51 every client that connects

0:55 and that

0:57 stream of request that comes from that

1:00 client is buffered and

1:04 read decrypted if it's an TLS and then

1:09 understood based on the chosen protocol.

1:13 So it could be HTTP2 HTTP11

1:15 or even just regular request uh

1:19 protocols such as SSH and other

1:22 protocols Postgress.

1:25 So then the back end after reads all

1:28 these requests it then processes them.

1:33 Optionally most of the time most backend

1:36 application need to connect to other

1:40 backend applications to consume

1:43 more data. Uh a classic example is an

1:47 API gateway or a reverse proxy where the

1:50 API gateway itself is a backend

1:52 application. Right? So a clients can

1:56 connect to the API gateway can send

1:58 requests but then that API gateway is

2:01 not the final destination you see right

2:03 so it turns around connects to some

2:06 other backend

2:08 so that's another socket connection but

2:10 it's a upstream socket connection and

2:12 then forwards that request to that

2:15 backend application so you see there is

2:19 always a front end socket connection and

2:21 then the backend socket connection

2:23 Another example most prominent is

2:28 the backend application server itself

2:32 may connect to a database server. So it

2:35 receives a request from an API gateway

2:39 or reverse proxy or load balancer. So

2:41 that's the front-end socket connection

2:43 and then it turns around and creates a

2:46 new socket connection.

2:49 Different protocol by the way.

The backbone of backend system design

Name: The backbone of backend system design
Channel: Hussein Nasser

Hussein Nasser

Key Takeaways

Start a conversation

The backbone of backend system design