Re: Is IETF is being shunned for new protocol development?

[Craig Partridge <[e-mail]>]

>
>
> Take a look at ESUN in OCP. They are developing a new network layer header
> for use in scale-up networks. The perceived problem is that IP is too much
> overhead. It's really the IP header plus UDP header since convention is
> that new transport protocols are encapsulated in UDP. There is a four byte
> header proposed that would be used instead of 28 or 48 bytes for IP v4/v6.
>

I've been watching but staying out of this discussion as it is as old as
IETF (it was an issue when I was on the first IESG) and I'm not sure I have
anything new to offer.

But as a protocol nerd, I couldn't let this technical statement go past...
We have been down this path of smaller headers/simpler headers/headers
multiple times in the past 50 years (cf. LOCUS, XTP, etc.).  They work in
tightly constrained environments and then fail miserably at wide area scale
and they show up at IETF and say "how can we layer this on top of TCP?"
[Meta point -- they almost inevitably want to scale to wider areas]

Those headers field in IP are there for very good reasons.  Yes many are
not needed in certain situations (a great way to see this is to look back
at TCP header compression - on a point-to-point in-order link you can get
TCP+IP headers down to, if I remember right, a byte).  But if you want a
fully general solution (and most folks who think they don't need a general
solution, cf. NetWare, eventually discover they do), there's not a lot of
wastage.

Thanks!

Craig

PS: Hugh Holbrooke, as my PhD student in the distant past, presumably is
aware.
-- 
*****
Craig Partridge's email account for professional society activities and
mailing lists.