Limiting the outgoing speed of the virtual interface

Question

linux

Hello!

There is a dedicated Linux server with one network interface and two IP addresses. Support for a second IP address is implemented using the eth0:1 virtual subinterface.

I would like to configure the server's web content to be displayed from eth0, and the distribution of files was carried out from eth0:1 using nginx, but the question arises - how to limit the channel width for eth0:1? Will tc work for this? Or does it not know how to work with virtual subinterfaces?

In general, please share your experience.

Answer 1

1. eth0:1 is not a virtual interface, but just an alias, it cannot have a separate queue discipline.
2. tc will do. Discipline is hung on the entire interface (eth0) and two child classes are created - one with a rate limit, the second - for the entire width. Send all packets with the specified sender address to the class with a rate limit (this will be the ip address on eth0:1). For details, I recommend contacting lartc.

Answer 2

By the way, you can limit the speed of content delivery and use nginx - nginx.org/ru/docs/http/ngx_http_core_module.html#limit_rate