Bandwidth - GigaACE Network vs Direct

jay2imagineit-net · 4 February 2017 20:09

This testing was one I intended to do long ago, but re-inspired from Neak’s post that included running Dante over the GigaACE link.

As A&H advertises it, the network connection between the MixRack and a Surface is a lower bandwidth link designed for management systems that may include Director, IP8’s, Wireless management and other systems that are now part of a regular show. It does this well, as long as you remember it has bandwidth limitations and latency. Audio does not travel well on it reliably, as noted below, but appears that it can be done within limits.

Observations of GigaACE :

Network bandwidth is shared with the console control of the MixRack. Over-utilization will cause the console to lag.
About 22% of the Gigabit bandwidth that you have vs a direct connect
Appears to not honor TOS/Prioritization, so latency-dependent traffic is subject to wild latency swings and dropouts due to network traffic between the surface and mixrack or other devices.
- Dante/Audio is directly impacted (tested, below). You can’t run reliable < 5ms latency.
- MIDI/rtp will be likewise impacted if it is being used for anything timing-sensitive (MTC, Clock)

Show recall timing (Util/Show Manager/Recall):

Nothing on network: 12.2sec
Full bandwidth test: ~44sec

If you have any questions on the testing I did, feel free to ask.

Jay

Environment:

Cisco SG300-28P Switch, VLAN 11, Configured for Dante QoS/TOS
MSS 1448 on all bandwidth tests
2 MacBook Pro’s. One direct to console, one via cisco switch at mixrack
dLive S-5000, DM64 on V1.40
Yamaha TIO 1608 for 2nd Dante device via switch at mixrack.
Using iperf3 -V -Z -t 600 -c 172.30.11.100


Direct Connect
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec   110 MBytes   925 Mbits/sec                  
[  4]   1.00-2.00   sec   112 MBytes   938 Mbits/sec                  
[  4]   2.00-3.00   sec   111 MBytes   931 Mbits/sec                  
[  4]   3.00-4.00   sec   112 MBytes   938 Mbits/sec                  
[  4]   4.00-5.00   sec   112 MBytes   938 Mbits/sec                  
[  4]   5.00-6.00   sec   112 MBytes   938 Mbits/sec                  
[  4]   6.00-7.00   sec   112 MBytes   938 Mbits/sec                  
[  4]   7.00-8.00   sec   112 MBytes   938 Mbits/sec                  

Via GigACE - no additional traffic load
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec  25.0 MBytes   210 Mbits/sec                  
[  4]   1.00-2.00   sec  24.8 MBytes   208 Mbits/sec                  
[  4]   2.00-3.00   sec  24.8 MBytes   208 Mbits/sec                  
[  4]   3.00-4.00   sec  24.8 MBytes   208 Mbits/sec                  
[  4]   4.00-5.00   sec  24.8 MBytes   208 Mbits/sec                  
[  4]   5.00-6.00   sec  24.8 MBytes   208 Mbits/sec                  
[  4]   6.00-7.00   sec  24.9 MBytes   208 Mbits/sec                  
[  4]   7.00-8.00   sec  24.8 MBytes   208 Mbits/sec                  

Via GigACE
1 Bundle of Dante - Audio Dropouts < 5ms
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec  16.4 MBytes   137 Mbits/sec                  
[  4]   1.00-2.00   sec  19.9 MBytes   167 Mbits/sec                  
[  4]   2.00-3.00   sec  22.1 MBytes   185 Mbits/sec                  
[  4]   3.00-4.00   sec  22.1 MBytes   186 Mbits/sec                  
[  4]   4.00-5.00   sec  22.1 MBytes   185 Mbits/sec                  
[  4]   5.00-6.00   sec  22.0 MBytes   185 Mbits/sec                  
[  4]   6.00-7.00   sec  22.1 MBytes   185 Mbits/sec                  
[  4]   7.00-8.00   sec  22.1 MBytes   185 Mbits/sec                  

Via GigACE
2 Bundles of Dante - Audio Dropouts < 5ms
[ ID] Interval           Transfer     Bandwidth
[  4]   0.00-1.00   sec  15.5 MBytes   130 Mbits/sec                  
[  4]   1.00-2.00   sec  19.1 MBytes   160 Mbits/sec                  
[  4]   2.00-3.00   sec  19.1 MBytes   160 Mbits/sec                  
[  4]   3.00-4.00   sec  19.1 MBytes   160 Mbits/sec                  
[  4]   4.00-5.00   sec  19.1 MBytes   160 Mbits/sec                  
[  4]   5.00-6.00   sec  19.1 MBytes   160 Mbits/sec                  
[  4]   6.00-7.00   sec  19.1 MBytes   160 Mbits/sec                  
[  4]   7.00-8.00   sec  19.0 MBytes   160 Mbits/sec

vilddyret · 5 February 2017 12:02

Only one question. Why would you want to do this?

jay2imagineit-net · 5 February 2017 15:21

Tor,

Do what specifically? Run the bandwidth test or run Dante over it?

I ran the tests to know if I was working with 10mbit, 100mbit or something else:

How many control devices can I safely put at the surface end? (Director, IP8, etc)
Could I bring up a remote screen session if necessary to control a laptop?
Does it interfere with the surface responsiveness?
Does the wifi router need to be at the rack or can it be at the surface? (Rack better)

Dante (or MIDI) wasn’t even a remote consideration until Neak mentioned he was doing it, so I included it to see how reliable it could be done. Saves a pair of ethernet runs if you only have a few bundles and can handle the latency.

I also have a desire to know as much as I can about something I am using that I depend on. In this case, the ethernet link “just worked” but the limitations were unknown. Also, I didn’t want to suddenly need it for something unexpected and not be sure it would work.

Jay

vilddyret · 5 February 2017 23:08

I am sorry, but I dont see any reason why you would want to bridge other data heavy network stuff through the GigaACE connection?? Both the Surface and Mixrack have dedicated Network ports for control (Director, IP8, etc), and you can always move the DANTE card wherever you want it, 3 slots in the mixrack, 3 slots in the surface. So why on earth would you want to contaminate the GigaACE connection? It is already handling +300x300 channels of audio, it is not made for handling other network data.

jay2imagineit-net · 6 February 2017 00:00

Tor,

I am not advocating heavy data - just putting forth what I observed based on comments from A&H (link below) and what I have heard in sales presentations. They present it as a usable network link between FOH and stage within reason.

https://community.allen-heath.com/forums/topic/network-port-uses

One thing to note, regardless of how much data you put through the network ports, you can’t interfere with the 300x300 audio data (see link).

For some people, they may not have the luxury of $1800 for an additional Dante card and adapter vs running a few channels back down the network. For others, it is nice to know you could run MIDI/MTC over it to sync a Virtual VTR on stage from qLab without an additional network cable required for a traveling show.

This was put forth for those that “have a need” to use the network, need to know the capabilities, but don’t necessarily have the time or ability to do the measurements themselves. That is all.

Jay

vilddyret · 6 February 2017 12:28

ok. I guess that could be useful to some, I would never do this myself, but hey, different strokes You can however interfere with the 300x300 link from what I understand. But yes, the mixrack will still work fine. I remember trying something similar on the iLive system, it was certainly not a succes. Audio coming out at the surface end can most definitely be affected.

neakmenter · 7 February 2017 10:24

Thanks for these figures Jay - It’s nice to get some prodding and poking done by some people so we can do some elimination of factors and variables if problems arise!

BTW - I’ve moved my Dante card down to the MixRack and ran a separate cat5e from FOH to see if that helped with my console lag. It didn’t seem to make any appreciable difference. I was mainly concerned that somehow the Dante was bleeding into the GigaAce connection (the very early revision of the dante card firmware had a bug that would cause audio to be pushed over the Control port). I do have the latest firmware but wanted to be sure, and to see if removing that Surface Port’s audio bandwidth from the single cable would help. It doesn’t seem to.

windsock · 21 July 2017 09:31

Hi Jay,

sorry to bring up the old post; but when you refer to a Dante “Bundle”, are you meaning a single channel of dante audio; a Dante “network flow” of 1-4 channels (as described in the dante training videos); or a single dante device with 30+ channels being transmitted or received?

thanks

jay2imagineit-net · 21 July 2017 22:03

Windsock,

A “network flow”, in Dante terms. Bundle comes from the days of Cobranet.

2 flows (8 channels) look good - beyond that you can really feel the impact.

Jay

windsock · 22 July 2017 06:07

Jay,

Thanks for the clarification. really helpful!

armchair · 24 July 2017 05:19

Jay et al.,

I want to thank you for bringing up this subject and your research. As a result it occurred to me to try to connect Powersoft’s control and monitoring application (Armonia) to the amplifiers over the GigaAce link. It appears to work just fine. I can have a laptop running Armonia (plugged into the general Ethernet jack at the console) monitoring all the amplifiers in the system (which are plugged into the general Ethernet jack on the Mixrack).

One caution for anyone wanting to try this is that the rear panel jacks on the K series amps use two of the 4 pairs in the Ethernet cable for AES3, so I make my connection to the Mixrack through the front panel jacks which leave those pins open.

I’ll continue to test but so far so good.

Best,

–Frank