rom: ietf <ietf-bounces@xxxxxxxx> on behalf of Stewart Bryant <stewart.bryant@xxxxxxxxx> Sent: 21 September 2023 10:37 On 20 Sep 2023, at 22:04, Brian E Carpenter <brian.e.carpenter@xxxxxxxxx> wrote: In terms of scope, it's always been pretty clear that the IETF is "above the hardware" (the main challenge being that MPLS is right on the boundary) and "up to generic applications" (the main challenge being the HTTP/HTML and URI boundary). Hi Brian Firstly MPLS is not a good example, it is really not fundamentally different from IP and the same forwarder usually does both*. It can be implemented fully in S/W, or in H/W is the same way that IP can. Normally both are implemented in microcode with hardware acceleration. As you get further up the stack you discover other features that are supported in hardware, for example there are hardware accelerators in servers for the transport protocols. Then of course there are features like NTP that require hardware support for optimum performance. Firewalls are not my area but I imagine they now have a lot of hardware assistance. Then we need to take a look at the mobile world where a bit stream is sent from the tower to a datacenter SDR and other than the use of accelerators this could be an entirely software system from radio all the way up to the application. So I don’t think the above the H/W scoping is correct in 2023. The questions is whether it was in reality ever correct? <tp> Generals are always said to be fighting the last war. My perception is that while I agree with the technical details, for me they are increasingly irrelevant. The landscape is dominated by a small number of large corporations who dictate what can and cannot happen. The bane of my life is the destruction of e-mail which impairs many things including my ability to contribute to the IETF. I link this to the rise of social media and the cloud. The problem is that we have little influence with those large corporations. In fact, at times we seem to be subservient to them, as with changing protocols to minimise round trips at the expense of other capabilities. Tom Petch Stewart * For good measure a lot of forwarders can do Ethernet bridge forwarding as well.
