Config for "multiplexing" non-caching proxy

[Chris Woodfield <rekoil@xxxxxxxxxxxxx>]

Hi,

I've noticed that either by design or as a side-effect of squid's  
caching that if I request the same object from multiple clients at the  
same time, squid will effectively "multiplex" the transfer - that is,  
use a single transfer from origin to feed the object to each client as  
it receives the incoming http transfer. IMO this is a Good Thing.

I'm wondering if it's possible to extend this behavior to a non- 
caching proxy - this would allow the utilization of squid as a non- 
caching CARP proxy while protecting a parent from large numbers of  
requests for a single object (which in a CARP setup would normally all  
get sent to the same parent, potentially overloading it).

So the goal is to utilize this multiplexing capability in squid while  
not actually caching the content. However, testing shows that when I  
disable disk caching (via "cache_dir null" or by "maximum_object_size  
0 KB", this behavior is no longer present.

What would one recommend as the minimum feasible caching config to  
enable this behavior, while either disabling or minimizing any  
"actual" object caching?

One option could be set up a small disk cache, but enforce a very  
short expiry time (say, 5 minutes) on all incoming objects. However, I  
don't see a refresh-pattern directive or other config option to  
enforce this on objects with longer Expires: or Cache-Control: max-age  
headers (override-expire enforces min-age, not max-age). Any  
suggestions here?

Thanks,

-Chris