[PLUG-AV] Questions about last night

[Harry McNally]

Hello AV list

Nick gave an overview of the outcomes of the WAIA broadcast and I'm going to 
pre-empt him this time and explain my confusion about the setup last night.

When Nick and I looked at the systems on Monday evening (UCC) I'd asked 
whether the network issues could be avoided by running an entirely separate 
(electrically isolated) local network for the AV system with a separate 
network connection to the interwebs so any disruption to internet broadcast 
did not impact on the rest of the local network; recording and communication 
could continue if uncontrolled network connections failed. I'd considered that 
the Puppet talk was the next test of what we were heading to for LCA.

Some work was done to confirm this was possible and Network Manager on the 
master and one camera slave was setup (with wireless initially) to confirm 
that NM was capable of this local network configuration. The plan was to move 
to ethernet local network and run that isolated system at the Puppet talk. I 
intended to get down and help Nick with that setup but got there late.

Here's how my brain (slowly) worked with the preconception of an isolated 
local network as I attempted to assist.

The decision to use the Spacecubed network was on the basis that "PLUG put it 
in and we know that it works". Unfortunately this assumption was an important 
experience from the POV of the LCA implementation that is very near.

AV = many other decision making voices

AV: "We don't want to have cables across the floor, just use the wall ports"

Me: "Hm. Ok so we can adjust patch cables on a panel somewhere to make our 
network ... ?"

AV: "No. We know this network works we don't need anything special here"

Me: <looks for wall ports and finds them fully occupied>

AV: "I think we can disconnect this black cable"

<Spacedcubed loses it's SIP phone for a while until we agree it's probably 
best to restore the cable ...>

AV: "This other cable goes to this access point"

Me: "Well, I'm not sure we can disconnect the wireless either ..."

AV: "It's ok there is a daisy-chain connection on the AP. Everything has this 
these days. We'll just connect to that"

Me: <thinks now> Oh. I'm really am behind the times but I don't understand how 
that is going to transparently manage our isolated traffic. I'm out of my 
depth here.

Further guidance meant the other camera went avslave->1GB switch->some office 
wireless router (of completely unknown config)->wall port->spacecubed switch 
(somewhere)->wall port->another switch of our own in the directors 
room->avmaster ...

My pondering of this later was that the network was installed (and like other 
homilies) "network use expands to consume all wall ports (and then people 
start adding their own switches for expansion ...)". I don't agree that the 
network was in a state that was predictable and I also roundly condemn 
everyone for not being able to read either Nick's or my mind and know what was 
expected (dammit)! ;-p <- Despite this emoticon, this was serious deficiency 
on my part of not articulating the decision we had made in isolation which 
other AV people needed to either accept/reject but at least understand what we 
were trying to achieve.

My observation and tentative suggestion is that at LCA/UWA if internal wiring 
is to be used (or cable runs added as needed) there will/may be wall ports 
that AV needed dedicated access to. There will/may be ports on patch panels 
that will/may need patch cables marked as part of the LCA AV network and rooms 
secured to prevent changes to those patches. There will/may be switches and 
routers that form part of the LCA network that AV may be required to pass 
traffic through. I'm assuming too that this woring must co-exist alonside UWA 
network and that they will be required to have isolation. There is much to do.

In addition, Luke Williams talk was prescient because I wonder if the 
configuration of these systems might be better managed with Puppet rather than 
imaging so they can be trimmed down from stock Ubuntu and any network failover 
config setup is well defined and reproducable. I'd made this suggestion to 
Jason last night and that perhaps Luke could provide some input. It was an off 
the cuff email comment but I think more than ever this morning that it is 
worth consideration.

Either way, I was happy to volunteer to assist with LCA AV but I am completely 
out of my depth with the performance and security that the networks and AV 
will require. I'm also happy to take instructions from an AV  director 
(including get out of the way!) and there will be others too who can help but 
(I think) we will just flap around helplessly if the systems are not robust.

Last night felt to me like a wake up call that some quiet urgency for 
acquiring components and adding to AV robustness prior to curtain call 
LCA2014. It is easy for me to say than contribute expertise so I apologise if 
I'm actually asking others for more.

My suggestion for enforcing documented equipment management into single 
"packages" or kits is a heartfelt one (sorry that is seems naff Luke). The 
equipment needs to be deployed with as much precision as you can muster and 
things going astray (as they have already) is a very simple and avoidable 
failure risk on the day.

More than my 2c
Harry