Sowmini.Varadhan at Sun.COM writes:
> I don't want to get too caught up in the nomenclature here,
> but my own initial instinct, like Jim's, was to club all 
> of it under ipadm. However, as becomes evident with property
> management in comparison with the other commands considered,
> it *is* true that the transport layer does not usually 
> target IP interfaces, just as the IP layer does not target 
> transport layer ports. It was for this reason that I split
> the tools into "ipadm" (or "netadm" or "l3adm"??) and 
> transadm (or "xadm" or "ulpadm" or "l4adm"??).

The existing parameters tend to be stack-wide globals.  There are a
few historical oddities to the contrary, but they're generally like
that.  The IP ones don't target interfaces and the TCP ones don't
target ports.

I think the question here is: what sort of granularity do we need (or
what objects will we operate on)?

If we're not going to try to tune individual blades of grass, a single
utility might well do it.  If we have to go very fine grained, then I
think "port" in one utility and "interface" in another may well be too
restrictive.  (E.g., setting a default MSS in TCP for a given remote
subnet, or an IP TOS value to use based on a TCP port.)

The other thing to consider is that if you're going to expose objects
like that, you may have only a short life before you run straight into
Crossbow's flowadm.

-- 
James Carlson, Solaris Networking              <james.d.carlson at sun.com>
Sun Microsystems / 35 Network Drive        71.232W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677

Reply via email to