On Mon, Mar 24, 2008 at 2:14 PM, Sunil Ghai sunilkrghai@gmail.com wrote:
On Mon, Mar 24, 2008 at 11:14 AM, Bruno Wolff III bruno@wolff.to wrote:
In case of dynamic throttling we won't be having any _fixed_ rate at
which
the connections assigned for updates will be able receive the
packets. It
means packets would be dropped frequently to implement policing.
Isn't this
waste of resources?
Yes and a protocol change was made to help. I believe this is the purpose of ECN (explicit congestion notification).
Tools like tc and tcng implement queues to control outbound data. Is
there
any similar _kind of_ option available for inbound data? (Obviously we can't have queues because once the packet has been
received
must be processed)
Shaping on the wrong side of a link is problematic. You can implement queues on the receiving side which might allow you to better control which flows get slowed down using IFBs (which replace the older IMQs). While you can't absolutely prevent the other side from swamping the link with low priority packets, things should work reasonably with well behaved applications.
Ingress shaping...sounds good! Basically it is the way to implement policing efficiently? -- Regards, Sunil Ghai
For traffic accepted on an interface, the *ingress* qdisc is traversed. It means all inbound data is traversed through it. So how do we differentiate as which inbound packet is for which application? port numbers? and who does it..operating system of *filters* attached with *ingress qdisc*?
If we want to implement policing on a particular connection, it's inbound packets may be dropped. But as *ingress qdisc* is common to an interface so how do we implement it?