LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

RE: L7 switching: string.patch for IPTables? (fwd)

To: <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: RE: L7 switching: string.patch for IPTables? (fwd)
From: Radu-Adrian Feurdean <raf@xxxxxxxx>
Date: Mon, 24 Sep 2001 22:36:20 +0200 (CEST)
On Mon, 24 Sep 2001, Zachariah Mully wrote:

> > Correct me if I'm wrong, but by the time you see the content in the
> > request fly by, the connection is already established.  By
> > that time, it's
> > too late, since you've _already_ done the load balancing yes?
>
>       Not if you're doing this on your firewall. In that case, you could make
> the decision to accept the connection and mark it to be passed on to the
> director. Even if you weren't able to do this on your firewall (stuck
> with a Pix or Nokia xxx), you could still do this on your director, i.e.
> the director/firewall combo... I am assuming though that packets will be
> first inspected by any iptables rules you have setup before being pass
> into the LVS layer. With fwmarks this would be a cheap and easy way to
> make a L7 lvs system.

No. LVS takes the decision at the first packet arrived (SYN/-, no data). When
the packet containing data arrives it is too late to take a load-balancing
decision. However you could do filtering this way - you wait until the
connection expires due to timeout - not very efficient.

However, this could work for UDP-based services, where there's no need to
establish a connection (at least at transport-layer level).

 Radu-Adrian Feurdean
mailto: raf @ chez.com
-------------------------------------------------------------
Majority: The quantity that distinguishes a crime from a law.



<Prev in Thread] Current Thread [Next in Thread>