Hi,

If I remember correctly, http v1.1 uses persistent connection to get
the objects referenced by an individual URL, and so it needs to keep
the same IP source for each URL request. This is an evolution of the
protocol, since it is not necessary to open a new TCP connection for
each http/get operation derived from the same URL request.

Load balance can be seen from at least 2 perspectives: the client and
the server. If your side is operating as a client, the usual is to keep
the same IP source for each connection-oriented request. If you intend
to load balance between your local web servers, for example, you can
NAT them and load balance each individual object requested by the
same URL, unless it is related to https (ssl) because the it needs
to keep track the connection history.

So, this load balance in a more general situation seems more like
a good project than a single task to be accomplished.

--
Antonio

On Thu, Feb 28, 2002 at 11:47:58AM -0600, Steve Westerhouse wrote:
> 
> I don't think it matters for web access since web access is stateless.
> Connections are created and dropped after each request.
> 

Reply via email to