Hi all,
I'm searching a CentOS based solution to have a control over the number of connections to various services depending on overall used bandwidth.
To give you a more accurate representation of what i mean, let's say we have some live WEB video feeds (M$ WME, embedded in WEB Pages hosted on an Apache server), a Darwin streaming on demand server (Links on WEB pages, Apache), FTP server and regular WEB traffic.
I'm already doing traffic shaping so the real time traffic has higher QoS (we use pfSense firewall). The problem, as you know, is that traffic shaping is not enough to ensure real time traffic will get through when bandwidth gets maxed out. The problem arise when the number of connections to streaming servers gets higher than the capacity of the link. I need to limit the number of connections to a safe maximum. The link is a symmetric 10 Mbps and we are not in a position to raise that for now.
I'd see something like a "traffic director" that sense the used bandwidth and controls the amount of connections to servers. Is there something that looks like that in CentOS? And how to inform the clients to connect later if their connection is temporarily refused (a timeout is not very elegant!)?
I heard that their are some commercial products that does about that but i'm searching for an open source solution if it exists.
Thanks!
Guy Boisvert IngTegration inc.