RE: iSCSI: More on naming and discovery

["John Hufferd/San Jose/IBM" <hufferd@us.ibm.com>]

David Black,
You said "...requiring a DNS resolution to find the storage required to get
the server up to the first point at which it can be worked on (e.g., single
user mode in Unix) is an invitation to trouble".

I am not sure that I believe that as much as you, however, there is
probably a reasonable compromise.

First, I think of our networks today depend on a DNS.  But also most of our
systems have a way to manually handle the inputting of an IP address. (They
do this because in some places, like Small-Office/Home-Office (SOHO)
networks there may not be a DNS.  But also there is the valid points that
you made about not being dependent on DNS.  In fact most vendors seem to
agree since they do not want their products stained with the problems
caused by a DNS.  (Yes, some IT groups do not have backups for their DNS.)
However, since most rational IT group have one or more backup DNSs, they
can usually be used.  I also know or situations where the vendor,
regardless of how the IP address was acquired, has made sure that they
recorded in some non volatile way,  the last IP address they used that was
in some way KEY to that application (or HW).  Sometimes this is in battery
backed RAM, sometimes in Eprom, etc.

I do think that from an "appliance" type mind set, we should factor in a
DNS approach where human understandable names are used.  It should,
however, be up to the vendor to decide if they have additional ways to
handle the first time startup IP address, and the saving of the last used
IP address, and even manual overrides.

Also there may be a valid point about using the DHCP technique for Boot,  I
think we need more discussion on that point.

.
.
.
John L. Hufferd