HTTP1.0/HTTP1.1, persistent connections and ident_lookup

A few questions relating to interoperability of HTTP1.0/HTTP1.1 Browsers, Squid
- current stable version - and
my associated Transparent Proxy firewall.

When answering, I'd appreciate it if you could make it clear whether any
unavailable features are either because no-one
has had the time to add the functionality to Squid, OR it is actually a
fundamental impossibility given the requirements for
compatibility and interworking of Squid and a variety of HTTP1.0/1.1 capable
clients and, possibly, parents.

At present, I have a rather ancient(!) version of Squid - v1.1.21-p5 running
internally, together with a mix of Netscape 3 ( mainly )
and Netscape 4 Browsers. I've noticed from the various release notes on the
Squid site, that more recent versions of Squid
support persistent connections - part of the HTTP1.1 protocol. The old Netscape
3 Browsers dont support this.
The Netscape 4 Browsers do.

The Squid cache itself has a "parent" - for want of a better term - of a
Transparent Proxy Firewall. Again the Firewall only supports
HTTP1.0, though I'm trying to get the manufacturers to support persistent
connections real soon now.

In the fullness of time, I'm going to get round to upgrading Squid. Similarly,
the Browsers are due for a massive upgrade soon.

If and as I reach the point that the Browers, Squid, AND the Firewall support
HTTP/1.1 at least to the extent of honouring persistent
connection requests from their respective clients, I'm wondering how much of the
chain can truly support this feature. I vaguely recall
a posting - probably from Hendrik Norstrom - indicating that although Squid
honours the HTTP/1.1 persistent connection request, it will
only issue HTTP/1.0 style non-persistent outbound requests - partially defeating
the usefulness of the feature.

Is the above true?

If it were possible for Squid to issue HTTP/1.1 requests, would this in any way
be undesirable?

If it is possible for Squid to issue HTTP/1.1 requests, is it possible to
'proxy' an HTTP/1.0 request from a client into an HTTP/1.1 outbound
request, and again would this have any benefits or drawbacks?

On another topic, the IDENT_LOOKUP feature in Squid presumably performs a lookup
request for every single inbound HTTP request, which
corresponds one-to-one with TCP connections for HTTP/1.0.
If I migrate to HTTP/1.1 capable Squid and Browsers, does this lookup still get
performed per HTTP request, or is the information cached
to the extent of making the lookup per client TCP connection, when I have
persistent connections in place? Thus hopefully saving on lots of small
TCP connections...

Ted Rule,
Flextech Television.
Received on Fri Apr 16 1999 - 04:22:25 MDT