[Tsvwg] Re: [e2e] e2e principle..where??....

John Day day at std.com
Wed Jun 6 10:52:50 PDT 2001

>On Tue, 5 Jun 2001, Narin Suphasindhu wrote:
>  > I don't care where I get the HTTP "objects" so long as I get them --
>  > and that they are reasonably accurate.  If my first attempt to get the
>  > object fails, I don't mind attempting to reload the object.  In fact,
>  > I think, HTTP proxy servers probably increase reliability in terms of
>  > making sure the clients receive the "reasonably accurate" objects.
>  > Whereas TCP connection splitting reduces e2e reliability -- because of
>  > the state-full nodes in between.
>  >
>also, in a lot of cases TCP connection splitting is implemented exactly
>in the form of a proxy(i.e at the application layer with sockets etc...)
>so TCP connection splitting maintains just about as much state as
>any other caching proxy...
>just as incorrect operation of TCP connection splitting can cause
>consistency problems..so can incorrect operation of caches...
>e.g. the cache can serve up stale pages..(if its not working correctly..)
>just the TCP connection splitting can cause loss of data if its not
>working correctly...
But TCP splitting has different semantics.  TCP is supposed to 
provide e2e reliability from the sourse application to the 
destination.  App -- TCP--TCP--App is clearly not doing that. whereas 
in the proxy case, a "query protocol" from Client to Cache and an 
"cache update" protocol from cache to website.  Two different things. 
The fact the query protocol from Client to Website is the same is 
purely coinincidental.

Take care,

More information about the end2end-interest mailing list