WEBVTT

00:01.240 --> 00:03.000
Hello everyone and welcome.

00:03.400 --> 00:09.840
In this video, we will learn about different transport types that MCP uses to communicate between client

00:09.840 --> 00:10.760
and servers.

00:11.240 --> 00:17.640
Here is the image that shows how the transport methods have evolved, offering increased flexibility

00:17.640 --> 00:18.920
and scalability.

00:19.560 --> 00:23.760
The very first one that we will talk about is Stdio transport.

00:24.200 --> 00:25.680
It's of type local.

00:25.960 --> 00:33.080
It uses basic systems like input and output streams, and it is used for running everything on the same

00:33.080 --> 00:33.720
machine.

00:34.280 --> 00:42.280
An example of stdio transport is command line tools like Bas scripts or Python CLI apps.

00:42.640 --> 00:46.360
The limitation of this transport type is that it only works locally.

00:46.640 --> 00:49.800
No network or cloud based communication is possible.

00:50.200 --> 00:53.320
The next one is HTTP with SSE.

00:53.840 --> 00:58.920
This is a significant evolution of HTTP over stdio transport.

00:59.280 --> 01:07.000
How this works is that it uses SSE for servers to stream responses back to the client, while it uses

01:07.040 --> 01:10.840
HTTP post for the clients to send messages to the server.

01:11.200 --> 01:15.960
The use case here is, you know, all the remote MCP servers.

01:16.520 --> 01:22.160
The limitation is that it requires an ongoing connection, harder to scale for many users.

01:22.600 --> 01:27.540
The third type of communication that happens is Streamable is HTTP.

01:28.220 --> 01:31.140
It's released in March 2025.

01:31.540 --> 01:35.100
It and it's a type modern cloud native transport.

01:35.420 --> 01:42.780
The way it works is it uses stateless communication and sends the request and receives the request in

01:42.780 --> 01:43.900
smaller chunks.

01:44.420 --> 01:50.540
The benefit of using Streamable HTTP is you no longer need long term connections.

01:50.940 --> 01:57.900
It is easier to deploy on serverless platforms, and it's great for modern and mobile applications.

01:58.260 --> 02:04.940
It is ideal for enterprise applications where you need scalable and low maintenance deployments.

02:05.380 --> 02:11.780
This would help enterprise applications work seamlessly and also scale seamlessly.

02:13.460 --> 02:15.020
So why does this matter?

02:15.420 --> 02:18.260
Why does this transport types matter?

02:18.700 --> 02:24.380
It actually helps understand how AI models interact with external environments.

02:24.820 --> 02:31.380
It also helps you understand to pick the right infrastructure and build the AI power tools that works

02:31.380 --> 02:32.980
best for your use case.

02:33.580 --> 02:40.180
It also reflects how softwares can scale from local testing to global deployments.

02:40.860 --> 02:47.700
I hope you learned in detail about these transport types, and I'll see you in the next video.