IPV6 is best for audio

[jabbr]

There you go. Credit goes to @plissken who has me convinced. No brain dead broadcasts. Discuss.

[AudioDoctor]

You can't just say that without proof and data, you know the rules.

[jabbr]

7 hours ago, AudioDoctor said:

You can't just say that without proof and data, you know the rules.

 

@plissken suggested this on another thread, and to keep that thread on its own topic (fiber) and after considering this, I want to elevate this question its own thread.

 

IPV6 does not depend on broadcasts and as we add little IoT widgets to our homes, the IPV4 broadcasts and might theoretically interfere with our music. VLANS can mitigate that (a switch can block the broadcasts from unwanted devices)

 

Ultimately once the connections are made, the packets are routed on the network using Ethernet frames. The IPV4 vs IPV6 addressing is mostly for device discovery essentially (and of course TCP layered on IP provides for packet reordering and retransmission ...)

 

So less broadcasting is generally a good thing.

[Miska]

9 hours ago, jabbr said:

There you go. Credit goes to @plissken who has me convinced. No brain dead broadcasts. Discuss.

 

Well, you wouldn't be using broadcasts for audio anyway.

 

But for example NAA has always supported both IPv4 and IPv6 and automatically prefers IPv6 over IPv4 when available.

 

In my opinion one of the benefits of IPv6 for audio is better QoS support.

 

[jabbr]

3 minutes ago, Miska said:

 

Well, you wouldn't be using broadcasts for audio anyway.

 

But for example NAA has always supported both IPv4 and IPv6 and automatically prefers IPv6 over IPv4 when available.

 

 

No of course not 😝 ... For years i've maintained an DHCP server on my Cisco firewall/gateway device. Spending a lot of time at home, I've had time to clean up a lot of "junk" and using avahi can connect using well known names. I think IPV6 works well/better with zero config.

[Miska]

IPv6 works better in almost all ways. Multicasts work better, QoS works better, no broadcasts (freaking SMB/CIFS protocol has been worst broadcast offender in IPv4).

 

Extra annoyance of IPv6 are the implementation differences between platforms, macOS can be very annoying for developer in that respect. IPv4 approaches have inherited more similar way to all OS from the old BSD Networking Release. IPv6 was implemented more in parallel on different platforms so it is not as consistent for the developer. The actual protocol works well though (thanks to good specs).

 

[plissken]

8 hours ago, AudioDoctor said:

You can't just say that without proof and data, you know the rules.

 

This isn't the objective forum.

[plissken]

18 minutes ago, Miska said:

IPv6 works better in almost all ways. Multicasts work better, QoS works better, no broadcasts (freaking SMB/CIFS protocol has been worst broadcast offender in IPv4).

 

Extra annoyance of IPv6 are the implementation differences between platforms, macOS can be very annoying for developer in that respect. IPv4 approaches have inherited more similar way to all OS from the old BSD Networking Release. IPv6 was implemented more in parallel on different platforms so it is not as consistent for the developer. The actual protocol works well though (thanks to good specs).

 

 

Not to mention more streamlined TCP header, linear subnetting vs IPv4 hierarchical, get to use FE80 addresses on routed links, Neighbor Discovery... The list goes on and on.

 

Unfortunately that long address freaks even my fellow engineers out that haven't taken the time to learn and I'm frikking stuck with IPv4 on a daily basis. Sigh.

[AudioDoctor]

26 minutes ago, plissken said:

 

This isn't the objective forum.

 

Its also the no sense of humor forum...

 

Left hand twist stranded wire sounds the best!!!!!!

 

That comment has as much veracity as yours however, without any reasoning or data to back it up.

[plissken]

48 minutes ago, AudioDoctor said:

That comment has as much veracity as yours however

 

Which comment was mine that you are referring to? As far as Data... When did you become an expert on IPv4 and IPv6?

[bluesman]

36 minutes ago, AudioDoctor said:

Left hand twist stranded wire sounds the best!!!!!!

That’s only true for the left channel(s).  Right twist is better for the right side, and untwisted strands sound best for center channels and subs.  Multiply the nominal impedance of the load in ohms by RMS power rating in watts to determine the number of strands.

 

I’ve validated this rule by conducting randomized, blinded A-B comparisons among copper wires with strand counts from 2 to 1000 driven by amplifiers with outputs from 5 to 1200 WRMS.  For all counts between 41 and 927 except 812 and 895, the above preference had p values below 0.05 with low and consistent inter-test and inter-rater variance.  The R-square for twist turns per inch vs p value was 0.7 in the left channels and 0.64 in the right.

 

We don’t need no stinkin’ sense of humor.

[jabbr]

10 hours ago, AudioDoctor said:

You can't just say that without proof and data, you know the rules.

 

Clearly this is a strawman  

 

The "data" is that IPV6 networks have less chatter (broadcasts). I don't expect hat once the music gets started that the audio bits will be significantly differently transmitted over Ethernet ... there are pros and cons of packet sizes that folks have gone back and forth on but the IPV6 has <1% overhead in that regard and other advantages ...

 

Whether broadcasts from other devices cause problems for music depends on your own network -- vlans can also mitigate if that's an issue, so generally I think that IPV6 is better and so far we don't have any reasons why its worse ... the goal here is a "best practice advisory"

 

I'm spending the day trying to reduce my own home network's use of IPV4 and learning about some idiosyncrasies of Avahi which I am using to let my network configure itself. For example: each of my interfaces  have 2 IPV6 addresses: one starts with "2002:" and the other (link local) with "FE80:". Curiously, my avahi-daemon (ipv6 only) serves up "2002" address at main interface e.g. "appserver.local" but when I want to specifically connect to my 100G interface, I have to place the "fe8o:" interface into /etc/avahi/hosts and that works as "appserver-100g.local" but if I put the "2002" address in, it doesn't work ...

[AudioDoctor]

33 minutes ago, plissken said:

When did you become an expert on IPv4 and IPv6?

 

AS soon as you became an expert on what's best.

[AudioDoctor]

21 minutes ago, bluesman said:

That’s only true for the left channel(s).  Right twist is better for the right side, and untwisted strands sound best for center channels and subs.  Multiply the nominal impedance of the load in ohms by RMS power rating in watts to determine the number of strands.

 

I’ve validated this rule by conducting randomized, blinded A-B comparisons among copper wires with strand counts from 2 to 1000 driven by amplifiers with outputs from 5 to 1200 WRMS.  For all counts between 41 and 927 except 812 and 895, the above preference had p values below 0.05 with low and consistent inter-test and inter-rater variance.  The R-square for twist turns per inch vs p value was 0.7 in the left channels and 0.64 in the right.

 

We don’t need no stinkin’ sense of humor.

 

At least someone here has a damned sense of humor left...  cheesus

[plissken]

2 minutes ago, AudioDoctor said:

 

AS soon as you became an expert on what's best.

 

You aren't making much sense. @Jabbr can you ask for and moderate / prune this thread? It has the makings of something informative if the SNR can be improved.

[jabbr]

5 minutes ago, AudioDoctor said:

 

AS soon as you became an expert on what's best.

 

@plissken *is* an expert in networking. That is his day job. @Miska is an expert in networked audio.

 

I don't have actual data that IPV6 is best for audio "SQ" because it currently uses a mixture. If you have something to constructively add, even your own impressions.

 

Sense of humor is welcome but that borders on personal attack, I mean questioning someones medical or surgical skills here would be equally offensive. Let's not devolve.

[bluesman]

45 minutes ago, AudioDoctor said:

At least someone here has a damned sense of humor left...  cheesus

I just couldn't resist that bait - but I was also hoping to decompress this discussion a bit because there's useful information in the thread & I'd like to know more.  Sadly, I now see the error of my ways and will quietly bone up on network configuration tools so I can reduce my current dependence on 4.

[plissken]

Here is a quick start to IPv6 address spaces:

 

1. If you see something starting in the 2000 it's public

2. If you see something starting in the FC it's private route-able

3. If you see FE it's link local or another way of saying TTL of 1

 

IPv6 are Base 16 (Hex-Decimal). Each place holder in an IPv6 address is 16 values 0-F (0-9, a,b,c,d,e,f).

 

If you take just a single octet #### that is a 16bit space and good for 65,353 addresses. There are 8 octets in an IPv6 address.

 

So you could have a private space of FC08:abab:acac:0001:0002:0003:0004:0005 and just that tail end represents the population of Bloomington IL.

[AudioDoctor]

47 minutes ago, jabbr said:

 

Sense of humor is welcome but that borders on personal attack, I mean questioning someones medical or surgical skills here would be equally offensive. Let's not devolve.

And if I start recommending best treatments online, please feel free to call me out. 

[AudioDoctor]

1 hour ago, plissken said:

 

You aren't making much sense. @Jabbr can you ask for and moderate / prune this thread? It has the makings of something informative if the SNR can be improved.

yes, just get rid of all dissent. I’m starting to see what the furor was about. 
 

the entire premise of this thread is flawed. Just because you and jabbr think something to be true, doesn’t make it so. 
 

that said, my initial reply was tongue in cheek. I’m sorry about the loss of your sense of humor. 

[jabbr]

2 minutes ago, plissken said:

Here is a quick start to IPv6 address spaces:

 

 

Got it, so why might

$avahi-resolve -n appserver.local

 

return a "2002" address

 

and I can 

$ssh [email protected] and it works ...

 

but if I put a "2002" address in /etc/avahi/hosts it doesn't work yet if I put a "fe8o" address in /etc/avahi/hosts then I can

 

$ssh [email protected] and it works?

[jabbr]

5 minutes ago, AudioDoctor said:

yes, just get rid of all dissent. I’m starting to see what the furor was about. 
 

the entire premise of this thread is flawed. Just because you and jabbr think something to be true, doesn’t make it so. 

 

No intention to get rid of your dissent, but on what basic do you conclude that the premise of the thread is flawed? Surely you can articulate a reason?

 

9 minutes ago, AudioDoctor said:

And if I start recommending best treatments online, please feel free to call me out. 

 

Barring what I learned in the 1980s, everything I've learned about networks, I've learned online... since we are conversing on a network, best network practices are expected to be delivered online ... not asking for an IETF RFC

[AudioDoctor]

5 minutes ago, jabbr said:

 

No intention to get rid of your dissent, but on what basic do you conclude that the premise of the thread is flawed? Surely you can articulate a reason?

 

how does that go, outrageous claims need outrageous proof?  Something like that anyway. It's 100% ok to "think" something is best but to just come out and declare it to be so without evidence and then argue about it?

 

I think I've heard that before and it usually comes from the guy yelling about how I'm going to hell while standing on his milk crate downtown...

[jabbr]

1 minute ago, AudioDoctor said:

 

how does that go, outrageous claims need outrageous proof?  Something like that anyway. It's 100% ok to "think" something is best but to just come out and declare it to be so without evidence?

 

what is outrageous about the possibility that IPV6 is best for audio?

[AudioDoctor]

Just now, jabbr said:

 

what is outrageous about the possibility that IPV6 is best for audio?

 

Nothing, you didn't suggest the possibility, you just declared it so, and then got argumentative about it when, jokingly I might add, called out.