[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Draft on Callout Protocol Requirements
I agree with Ian. An OPES service needs to be explicitely requested and must
not brake the end-to-end connection. The place to do this is on proxies the
client connects to, where the connections are terminated.
I don't see OPES services working out of an L3 sniffer kind of device.
Christian
> -----Original Message-----
> From: Ian Cooper [mailto:ian@xxxxxxxxxxxxxxx]
> Sent: Wednesday, November 28, 2001 11:22 AM
> To: ietf-openproxy@xxxxxxx
> Cc: Phil Rzewski; Reinaldo Penno
> Subject: RE: Draft on Callout Protocol Requirements
>
>
>
> OK, it looks like the confusion here goes back to assumptions
> of what the
> phrase "IP connectivity" means - something I decided to not
> comment on
> earlier.
>
> The point seems to be that you either have a content network in place
> before you overlay an OPES environment on it, or that you
> install a content
> network by installing an OPES environment box or boxes (the
> confusion here
> can probably summarized as "what is OPES?"). In the latter
> case, so far as
> you "only need IP connectivity" to do this, you're pretty
> much right. But
> I don't think you can have OPES services without having some
> application
> layer intermediary/host on which to run them. For example,
> my initial take
> on the "only need IP connectivity" point suggested you might
> be thinking of
> putting OPES services in something like a (L3) router...
> which is a big
> no-no.
>
> --On Tuesday, November 27, 2001 16:26 -0800 Phil Rzewski
> <philr@xxxxxxxxxxx> wrote:
>
> >
> > At 03:22 PM 11/27/2001 -0800, Reinaldo Penno wrote:
> >
> >
> >>
> >> > -----Original Message-----
> >> > From: Phil Rzewski
> >> > [<mailto:philr@xxxxxxxxxxx>mailto:philr@xxxxxxxxxxx]
> Sent: Monday,
> >> > November 26, 2001 11:34 AM
> >> > To: Penno, Reinaldo [SC9:T327:EXCH]
> >> > Cc: ietf-openproxy@xxxxxxx
> >> > Subject: RE: Draft on Callout Protocol Requirements
> >> >
> >> >
> >> <snip>
> >>
> >> >
> >> > In brief, we know that in the routed world, once you put two
> >> > host machines on
> >> > the same ethernet wire, you technically have a "network". As
> >> > "networks" go,
> >> > it's on a whole different scale than one operated by AT&T or
> >> > the IT department
> >> > at your local enterprise, but this is why we have terms like
> >> > "public backbone
> >> > network", "LAN", "VPN", etc. Similarly, we would argue that
> >> > once you add ANY
> >> > proxy ("forward" or "reverse") you have created a type of
> >> > Content Network.
> >>
> >> okay, I can agree with that since it actually adds to my
> point that to
> >> have OPES you only need
> >> IP connectivity, nothing else. You do not need an CN (and
> not by a long
> >> shot a CDN) overlay network in place BEFORE you install
> OPES devices.
> >
> >
> > I'm not sure how you get that from what I said, though
> maybe I'm failing
> > to understand something. Back when I was first tracking OPES, the
> > services were assumed to be provided by either proxylets or
> callouts to
> > external boxes. The device that was making those callouts or calling
> > those proxylets was typically described as being a
> surrogate or proxy of
> > some kind. If that's still true (is it not?), then by my
> definition, you
> > WOULD have a Content Network before you layer OPES
> proxylets/callouts on
> > top of it.
> >
> >>
> >> You end up having a CN (not an CDN) overlay network AFTER
> you install an
> > OPES
> >> device.
> >
> >
> > This I'd agree with, but I'm agreeing with the literal statement. :)
> > That is, I'd say that any network that contains OPES services is a
> > Content Network (I think it had to be before you added the
> OPES services,
> > even). It may be a CDN, since a CDN is just a specific type
> of Content
> > Network. That's determined by whether it contains components like
> > distribution, request routing, etc.
> >
> > --
> > Phil Rzewski - Senior Architect - Inktomi Corporation
> > 650-653-2487 (office) - 650-303-3790 (cell) - 650-653-1848 (fax)
> >
>
>