#systemdesign #softwarearchitect #coding

追加: 2026-05-06

[00:00:00]Well, it's a pretty valid question and in a 60 seconds I'll tell you how to answer this. So, this is day seven of 30 days 30 interview questions where I talk about frequently asked question by big tech companies. So, in the last we talked about what is the difference between HTTP 1 and HTTP2 and how HTTP2 is so fast by using just single channel.

[00:00:15]But here's a catch. Suppose you are on Apache network for example a metro and you are using a website that you know in turns makes 30 requests for CSJS files and everything. So if you are using HTTP2 it will be using multiplexing. So all the request and the responses that needs to go will be broken down into smaller frames and they'll be traveling in this one TCP connection. So it is really fast but the problem with this is it is still using TCP and if you remember if a packet drops in TCP everything else is stopped until that lost packet is retransmitted. So just one drop and everything frozen. On the other hand if you were using HTTP1 you would be using six different channels. I know the request have to travel sequentially but if one channel is blocked it is not blocking the other channels. Other channels are free to move. So you're not totally frozen. The damage is isolated. on key network HTTP1 is still the better choice. Second reason is half the internet still run on HTTP1. You can't just force upgrade them all. So we have to keep it for fallback.

[00:01:01]There's one more reason and I have given that in the pin comment. So do check it.

[00:01:04]Until next time. Thank W.