Brave New World: remote backup rig via lightpipe split

[Dave Labrecque]

Jazz Trio this weekend at a new PAC South of town. Here's my plan:

Mic split into ADA8000s
ADA8000s out to lightpipe, which will be split to both...
My standard Digiface/laptop rig, and...
A currently-coming-together backup rig (older PIII tower) using a second, rented Digiface (thank you Mark Stebbeds)

My thought was to slave the ADA's off of the main Digiface, then slave the backup Digiface off the ADA's (all via lightpipe). Then it occurred to me that if the main system goes down, depending how hard it goes down, there might be a chance that the master clock stream from that Digiface will hiccup, effecting the downstream ADAs and backup Digiface. The solution in that case would be to make the ADAs master and both Digifaces slaves, so that if either computer crashes, the other would be safe.

But that means using one of the ADA's clock as master, and I've always had the impression that the RME clock is superior. So, I'm wondering if there's an option I'm overlooking (using the hardware outlined; no, I don't wanna go buy a Big Ben. ).

Thoughts?

[DominicPerry]

OK, here's how it goes in IT land.

1. If you want a fully redundant system, it's going to cost you more than twice the price. This is because you need two of everything, plus some extra bits which make it all hang together.
2. It's going to be more complicated because you are going to have more than twice as many components.
3. It's more likely to fail because you have more than twice as many components.
4. Failover will almost never be seamless.
5. During testing, failover will appear to work without manual intervention. In practice, failover will fail to failover and will require manual intervention.

Trust me. I did this with many systems, from Novell Netware, to Oracle databases, to Oracle ERP clusters, to VMWare virtual machines, to international networks.

What's my point? Two main things.

1. Just to tell you that it's really difficult and it needs a lot of methodical testing. You need to break every component in turn and see how it fails over. Yes, that includes the backup devices, which sit there only for backup, but now because they're in the mix, if one of them fails, it can still bring the whole thing to its knees.

2. You need to think about the reliability of the individual components. The tendency in IT land is to always have two (or four) of everything. This is absurd. You need to know what is likely to fail. Base this on what has failed to your knowledge in the past. Has your Digiface ever failed? Have you heard of anyone else's Digiface failing? Has a Behringer ever failed? Has your Behringer failed? So which device do you want to trust to be your master clock? Digiface. No question.
What is the most likely failure? Power. Always power.
Realistically, as the ADA8000s are the least reliable component, you should have them redundant. Which means mic splitters. Seems like you've got that covered.
Power the Digiface separately from the laptop. Then kill the laptop in 17 ways. Pull the power. Hibernate. Go into standby. Play a game on it. Crash SAW by loading it up with 300 plugs. Pull cables. Pull all the ADA8000 lightpipes from it at the same time. See if the laptop can fail and the Digiface can keep providing a master to the other Digiface. If so, I think that is the way to do it. You have a problem, which is that wordclock doesn't have a protocol to keep two devices in sync, with one taking over when the other fails. There's no getting round it. So you have to make the Master clock as bomb proof as possible.

Hope this helps.

Dominic

[Bill Park]

Wouldn't it be cheaper and safer to split the mics? That is what we do, anyway. 10 channels of 3x split rent for about $25 around here, but I'm sure that some of the rock rental houses have splitter snakes with 24-56 channels that would rent for $60-$80 a night.

Bill

[ffarrell]

At a Political Action Committee?


mark

Performing Art Centers, sometimes ran by what you said.

fvf

[Dave Labrecque]

Thanks for your thoughts, Dominic. I appreciate tapping into your years of experience with redundancy in IT.

My thought is that a computer is much more likely to fail than an ADA8000. That's why my redundancy is there. I've never had or heard of an ADA8000 failure. I've heard of and experienced computer failures many times. Once during a remote recording gig.

So my question about which to use as master was more about the inherent "sound" of the clock than about failure of the unit that's running it.

I hope I have time to test out crashing the master Digiface computer to see what it does to the slave stuff. Taxes are killing me this week.

[Dave Labrecque]

Wouldn't it be cheaper and safer to split the mics? That is what we do, anyway. 10 channels of 3x split rent for about $25 around here, but I'm sure that some of the rock rental houses have splitter snakes with 24-56 channels that would rent for $60-$80 a night.

Bill

Hi Bill,

Safer, yes. Cheaper -- I don't think so. It's backup DAW (my way) vs. mic splitter plus backup DAW (your way). Unless I'm misunderstanding you.

Tucson being a small town there aren't a lot of mic splitter options afoot. I get one for $60 from the local big-time live sound company. They only have a couple of options.

Still might be worth it, though, thanks.

Mark -- good point on eliminating the crash/clocking issue if I double up on the Behringers, too.

[Dave Labrecque]

Or if you only need eight channels as I thought you said, just move unit one over to the other rig. Or maybe I misunderstood.

Mark

Yeah, that's what I mean. I could even set up a 2nd backup system, since I have three Behringers.

[Cary B. Cornett]

I've never had or heard of an ADA8000 failure.

I never had one fail during a gig, but I did have one fail after being allowed to "rest" for a few months. Luckily, it was just barely within the warranty. Later, when I bought another one, it was DOA and I had to take it back for replacement. That's 2 out of 4 for failure rate.

[Dave Labrecque]

I never had one fail during a gig, but I did have one fail after being allowed to "rest" for a few months. Luckily, it was just barely within the warranty. Later, when I bought another one, it was DOA and I had to take it back for replacement. That's 2 out of 4 for failure rate.

Well, maybe you think this is splitting hairs, but to me failure on the job is way different than DOA, and a bit different than after "resting" for a few months. Not to say that these happenings don't have some importance in the overall reliability of the line.

Also, you may consider a larger sample size before deciding on 50% failure rate. I don't think I've heard of any other actual failures of any kind.

Definitely too bad you're the guy who got the bad luck.

[Wink0r]

Well, maybe you think this is splitting hairs, but to me failure on the job is way different than DOA

They coincided one time for me - - - 2000 miles from here. I took two new amps to Sturgis and one almost came on. Luckily I was carrying a spare amp, so the event went on.