[lpi-discuss] IPv6 in exam LPI ?

[Bryan J. Smith]

Etienne Goyer <etienne.goyer at videotron.ca> wrote:
> I think IPv6 is definitely *outside* the scope of LPI,

I disagree.

A system administrator needs to know not only what is a
proper IPv6 host/mask looks like -- for at least "ifconfig,"
if not "route" -- but at least needs to know which IPv6
address is the local, layer-2 physical LAN compare to the
other 1-2 addresses that are not.

It's like asking about the LINKLOCAL IPv4 169.254.x.x subnet
when you don't get a DHCP address.  The concept of LINKLOCAL
for IPv4 actually came from the fact that there is a
dedicated LINKLOCAL in IPv6, where the bottom 64-bit bits
always match in IPv6 (based on MAC address in most cases),
and there is 3 bits near the front header that designates its
the local LAN.

> and it is too scarcely deployed to get any value in having
> it tested.

Distros are coming with IPv6 enabled on systems, SuSE most
notably.  Understand the LINKLOCAL IPv6 address is _always_
the same for a IPv6 node -- a 3-bit network header plus the
last 64-bits being based on the 48-bit MAC Address.  Since
this is coming on distros, people should know it.

They should also be able to identify a IPv6 address, and all
its short forms.  That way they aren't confused that it's a
MAC address or IPv6 address.

> In 10 years, maybe.

I'd argue the next LPIC-2 rev.  I'd almost want to push for
"basic identification" of what an IPv6 address is, and the
fact that _every_ IPv6 node has a LINKLOCAL address in
LPIC-1.  Again, distros like SuSE are coming with the IPv6
LINKLOCAL up on default installs.


-- 
Bryan J. Smith                | Sent from Yahoo Mail
mailto:b.j.smith at ieee.org     |  (please excuse any
http://thebs413.blogspot.com/ |   missing headers)