[2911] in North American Network Operators' Group
Re: topological "closeness" (RE: Web o' Wonder)
daemon@ATHENA.MIT.EDU (Jonathan Heiliger)
Mon May 13 18:46:19 1996
Date: Mon, 13 May 1996 15:39:50 -0700 (PDT)
From: Jonathan Heiliger <loco@MFST.COM>
To: Mike Trest <trest@atmnet.net>
Cc: nanog@merit.edu
In-Reply-To: <2.2.32.19960513194819.00692964@pop3.atmnet.net>
On Mon, 13 May 1996, Mike Trest wrote:
|} Yes. Initial customer reaction is, at first, hesitant. However, when
|} they try it an see *DRAMTIC* improvements in response times. This is
|} predicated on the "closeness" of the cache relative to dragging the
|} stuff over the whole world.
There is definitely a benefit to the customer once the data is stored on
the cache. However, in order to scale at a reasonable cost one would
probably want to build a hierarchy in the caching servers.
With or without a hierarchy of caches (such as local area and network
border caches); the first user to request data through the cache will be
penalized with an additional delay not present in direct access.
-jh-
