D&C GLug - Home Page

[ Date Index ] [ Thread Index ] [ <= Previous by date / thread ] [ Next by date / thread => ]

Re: [LUG] MAC address on Android


On 14/01/14 09:53, Philip Hudson wrote:
> On 14 January 2014 09:25, Tom <madtom1999@xxxxxxxxxxxxxx> wrote:
>> And you cant see the downside of that at all?
> Actually, I can. You see, I've not only used this stuff, I've administered
> this stuff. A lot. Didn't say there was no downside, but I do say that for
> both the user and the administrator there is no comparison in terms of
> getting rid of utterly unproductive, unnecessary, annoying, distracting,
> special-knowledge network configuration tasks that are utterly opaque and
> unusable to naive, low-skilled users and that require a highly paid
> specialist to understand and resolve. (Could *that* be the real appeal of
> DHCP?) It's the difference between designing a system that can fail *in
> ordinary usage* and one that can't. Like designing a system that can have
> either sound or graphics or both either working or not working and still be
> considered unbroken, and one that either has them both working or is broken.
> So, yes, NBP's downsides exist. I'll let you have a go at listing what
> you've spotted and I'll tell you if you miss anything. But DHCP is still
> *rubbish* tech.

You're both right and wrong here: DHCP *is* pretty rubbish to be fair, I
can't really argue with that. I still wish that there was some kind of
equivalent to MX priority for DHCP, as anyone who has ever had more than
one DHCP server on the same subnet will also. Managing them is also a
bit of a pain in the arse, especially when they're being misused for
things like asset management via spreadsheets full of the MAC addresses
of the 2000 new Dell desktops being dropped off on Monday.

But, networking is hard, especially if you're doing it right. IPv6 will
eventually take most of the pain away, when lazy ISPs, hardware
manufacturers and the rest of the world finally catches up. The problem
is that IPv6 *is* hard: IPv4 is already rocket science to non-computer
experts with all of it's gibberish about subnet masking and default
routes, and IPv6 is utterly impenetrable to anyone except network and
systems administrators. Which is convenient for me :]

Where you're so, so very wrong is praising Apple for, of all things, the
pile of sh*t that passes for network functionality in their OS. I mean,
you're kidding right? I too have done admin for Apple machines, and a
lot of it. On your brain-dead 2 or 3 machine, autoconfigured default
home subnet behind a SOHO NAT router they work fine, if
noisily. Usually. On a managed network? Exercise in pure pain. As
someone else said, there's a reason why you stick the graphics
department's handful of Macs into their own subnet or VLAN, feed a bunch
of highly restrictive rules to pf on the OpenBSD router that stops them
from destroying themselves and the rest of the network and call it a
day. Just like IPX, DecTalk and NetBEUI there is a reason why AppleTalk
has gone away and died - because it was hopelessly bad. Bonjour and
Zeroconf continue to annoy me to this day unfortunately. In the brave
new world of OS X 10.9, aka, Mavericks, Apple are bizarrely incapable of
allowing two different subnets to talk to each other over a L2TP VPN if
both sides are behind a NAT gateway for example. Please never use the
words "Apple" and "networking" in a positive light in the same sentence
again. Also note they are finally deprecating their idiotic AFP rubbish
in favour of Microsoft's far superior SMB protocol as they've evidently
finally realised what a bunch of morons they are. And Apple have broken
NFS compatibility between their stupid OS and standard Linux/UNIX boxes
than AIX has, and that's really saying something.

Not that I'm bitter or anything. There is a reason why I chose "bad
apple", and a not insignificant part of it involves networking.


The Mailing List for the Devon & Cornwall LUG
FAQ: http://www.dcglug.org.uk/listfaq