question about keep-alive

[imba-tjd]

Hi. As far as I know, browser uses the Keep-Alive mechanism by default for HTTP/1.1. Why do you say In real, user doesn't query same URL using Keep-Alive?

[hatoo]

I assume oha is used to benchmark a single URL (although -rand-regex-url feature is implemented)
And by disabling Keep-Alive, we can mimic a situation where many browsers access the same URL.

Benchmarking a single URL with Keep-Alive is like a

GET / HTTP/1.1
...
GET / HTTP/1.1 # in the same connection, it is unusual since you must know the content of it already
...
GET / HTTP/1.1
...

It is something weird.

[imba-tjd]

Thanks. I think i know. I just read the article about "coordinated-omission"

The goal for benchmark tool is to simulate many users connecting at the same time, rather than a single user sending many request. Disable Keep-Alive can achieve this.

But I think the wording in readme needs changing. Or maybe when someone else has this question, give them a link to this this discussion.