Almost everyone has asked themselves what would happen if the system went down. There are many different approaches out there and many basic thoughts have already been discussed:
Automatic switching between two completely separated identical systems like the techs of the stars do it is the top end of the line. Even heard about the insane idea of cloning the hard drive prior to each performance to have everything 100 % identical
Others just have a spare machine they fire up if something crashes
Some backup only the files
Some don’t have a backup at all
The funny thing is that everyone seems happy with their solution. So it’s just up to you to decide based on budget, weight, and the switchover speed.
I’d like to focus on the thoughts that led me to a solution somewhere in between the ones mentioned that I’m pretty happy with.
Have I ever needed it? More on that later…
The requirements
I feel too unimportant to run two identical setups, but there’s one thing I knew I didn’t want any longer: a “low-budget” version of my sounds programmed on an old hardware sound module - it has to be the same gig file for whichever I use.
Also the switching has to be instantaneous.
Main decision
The main decision was to have a hot backup only for everything that is software, meaning that I have two Macs sharing the same connected hardware.
(I also have a backup plan for the hardware part, but it doesn’t work instantaneously.)
Hardware
My decision requires to have devices that allow for two hosts.
iConnectivity does that really well so I chose the mioXM for my MIDI interface which is connected with USB (Main) and LAN (Backup).
The use of AVB across our stage basically limits the suppliers of audio interfaces for my 7 audio channels to MOTU and Presonus. Even more if you also want parallel analog outputs, too, for rehearsing and so on. The main Mac connects via USB, the backup directly via LAN (AVB).
I went for the Presonus 16R, a rackmount 1 U digital mixer, because it seamlessly handles hot plugging of USB and AVB without any syncing issues or dropouts. In contrast, the MOTU UltraLite AVB went completely silent for many seconds when either USB or AVB is plugged in or out or one is restarting.
If you go without AVB, iConnectivity again could be a good solution also for the audio part.
Last, also the OSC controller (TouchOSC) needs to be able to deal with two parallel instances of Gig Performer. While this is mostly solved with programming, my requirement of having a wired connection involves just a little bit more thinking.
Next, we’ll discuss what this approach works in an emergency, how the gig file needs to be adapted and how to keep both in sync.
From my last 30+ years: I had only one evening in the 90ies where my masterkeyboard failed, so I had to finish the gig on my M1 only
I think the most important thing is to select and maintain the ‘right’ reliable equipment, with proper cabling etc.
Of course with the rise of mac/PC based rigs the risk is slighly higher that a central piece may fail or need a restart etc.
Therefore I still have not completely moved to midi controller keyboards only - I still could use my Kurzweil to play most parts of a gig with some limitations.
Otherwise: have enough spare cables, mics, a spare sustain/switch pedal etc. and you are fine to enjoy nearly any gig
BBB
PS: a UPS is meanwhile a basic component of our bands rig to keep basic things like keys, wireless, mixer etc. up and running when mains power gets in trouble.
PPS: look at your drummers/guitarists: do they pack additional drums or amps? I don’t think so
I employ a second instance on my main computer with predictive loading enabled as a quick backup in the very rare instance of a crash (always related to a plugin and not GP)
Since it’s using predictive loading, it’s not adding much to the load on the computer and I can be up and running in seconds while the main instance restarts.
I also highly recommend a UPS for any setup. Too easy to get a small power drop which can cause a computer to restart or lock up (I run a Mac Studio, so no laptop battery to protect against even the shortest power loss)
I also pack a spare, lower spec’d Mac mini with a recent backup as an emergency backup in the case of hardware failure or damage
You’re right, @tripleB , that with reliable equipment, which may include both hardware (cables…) but also software (plugins…) the need for an emergency can be lowered.
I think to single most underrated need for a backup is the bass drum pedal! A bass guitar can be routed directly into the mixer, a guitarist can carry a small amp simulation. But I guess the expectations are higher for keyboards - we get the angry looks if it’s not the exact same sound every time.
It’s funny that with all my thoughts about backup plans, I have never felt the need for a UPS yet. OK, my main computer is a Laptop, but I always figured that there are so many pieces in the entire chain not backed up by a battery that I wouldn’t be too much of a limiting factor until my system is back running.
But it’s definitely worth considering. I bet they have become much smaller since the last time I looked for one.
Also very interesting is a second instance of Gig Performer being the first backup, @amosdef, thought about it all night!
I might not understand your point correctly, but predictive loading doesn’t generally reduce the “load on the computer” (CPU). It only speeds up the loading of the gig file and lowers the RAM usage.
But even for RAM, something strange happens. I loaded my gig file in two instances in parallel. The activity monitor shows the
Individual usage of RAM:
2,5 GB (instance with predictive loading) + 18 GB (instance without predictive loading)
18 GB of RAM for each instance (both without predictive loading)
Total usage of RAM:
28 GB (one instance with predictive loading, the other without)
28 GB (both instances without predictive loading)
Obviously two identical gig files do not add up to double the total RAM usage so there must be some sharing going on in the background. This indicates that there might be no advantage of running your parallel second instance with predictive loading.
I used that term loosely with the assumption that predictive loading would save some resources (mostly RAM) with the second instance. I’d have to test this theory out when I’m at my production computer again.
My #1 reason for using the second instance is for a sort of hot backup type of solution within the same computer.
Since I’m using the Mac Studio, I have some extra headroom so the extra resources used was not a major concern. I also prioritize physical modeling plugins that are well written and mostly only use CPU power when in use, even when predictive loading is not used, so having a second instance doesn’t seem to cause any issues, but everyone’s setup will be different so my “solution” might not apply to others
