Servers - a complicated matter
Posted by: Jonas Olofsson on 22 July 2016
While streaming opens up a fantastic way of enjoying music, servers seems to add a bit of complexity to the equation.
To be able to produce a server, working for a long time with out any problems, and, if problems show up, being able to repair, is a complicated machine.
I do think audio business in general isn't set up for that and that serious server producers with a heart for audio is what you have to look for over time.
Personally I wouldn't go back to CD or LP since the conviens is to big.
So, rebility could be first choice
//Jonas
And it gets trashed by my vinyl edition of The Beatles in mono somehow though the vinyls won't fit on my NAS - bays are too small
Adam Zielinski posted:And it gets trashed by my vinyl edition of The Beatles in mono somehow though the vinyls won't fit on my NAS - bays are too small
Ah, The Beatles. I still have my 1968 Mono original, a Stereo version, several CD versions and the 2009 remastered version. Last record player I had was a Linn Axis which I may or may not still have but certainly never used in this century. ��
andarkian posted:Adam Zielinski posted:And it gets trashed by my vinyl edition of The Beatles in mono somehow though the vinyls won't fit on my NAS - bays are too small
Ah, The Beatles. I still have my 1968 Mono original, a Stereo version, several CD versions and the 2009 remastered version. Last record player I had was a Linn Axis which I may or may not still have but certainly never used in this century. ��
The Beatles 2009 Remasters. The data is on my server but this thing sits on my Fraim:
I believe media servers for the majority can be very simple affairs, and in most cases residing on a little NAS box.
Now as far as SQ differences between servers I have been and continue to analysis here, BUT so far I have found an apparent correlation between server media transfer behaviour and subjective SQ.
I have found media servers (software / OS / hardware combination) that have the lowest and most consistent inter TCP packet timing variation sound the best. By sounding best I mean really quite marked in terms of presence, timing, and sheer naturalness with PCM wav.
I have found when using Asset or MinimServer on my servers and transcoding, the TCP timing variation increases, and the SQ starts to relatively fall away.
I have configured some custom filters on WireShark to capture and illustrate this graphically using a mirrored port on my switch connecting to my Naim streamer.
The above analysis on my setups so far has led me to use for best SQ a simple DLNA media server that resides on a low powered NAS (Netgear NAS 102) with NO transcoding streaming PCM WAV for optimum SQ.
There it appears simplicity is best, and my technical analysis does seem to illustrate this.
i may be able to use DSCP packet markings to regulate and recover the minimum TCP timing variations on a QoS enabled managed switch... that is now on the todo list.
Simon
Simon,
I'm afraid I don't understand the technical side of what you are describing, but I understand you are saying 'keep it simple' in essence. Something I find interesting is using my Synology NAS, simply switching between Synology's own media server and Minimserver, I find a tangible difference where I prefer the SQ of Minimserver! This is using .wav files stored on it, with no transcoding.
Seems strange that there should be a difference?
Jason - when I hear a difference and people reporting x sounds better than y - sometimes its chatter, noise and self justification - but other times when you see a correlation there might be something in it. This is what I have attempted to look at and better understand; by removing the clouds of marketing and sales obfuscation and sometimes over simplification and look underneath the covers at some of the engineering and technical parameters of streamed audio.
My initial findings have shown a correlation of certain network data 'dynamics' with SQ. This sort of thing is not necessarily unusual or strange in my professional world - so I have applied some of my thinking of that world to the world of domestic audio replay systems.
So far I have been able to replicate and and analyse to confirm the likely contributory causes of some of the best SQ I have heard. To date some of the best SQ has been achieved by using a very simple modestly powered NAS and combined DLNA media server simply offering Wav PCM files without transcoding from FLAC.
So far I have been able to reasonably well predict a data flow pattern based on the heard SQ. More to do..
Simon
Jason posted:Something I find interesting is using my Synology NAS, simply switching between Synology's own media server and Minimserver, I find a tangible difference where I prefer the SQ of Minimserver! This is using .wav files stored on it, with no transcoding.
I've tried this a number of times & have not detected any change in SQ that I can conclude as either different or better.
Using my NDX all wired via a switch from Synology DS214 (CPU Dual Core 1.066 GHz & RAM 512 MB DDR3) I've used FLAC straight & transcoded & WAV with both Syn's native Media Server & Minimserver.
Media Server is rubbish with transcoding, but as I'm all WAV & DSD, I tried it for test purposes only.
With WAV the only difference I can detect is Minimserver uses more RAM & CPU than does Media Server.
The upsides with Minimserver are transcoding, plus it does a better job of art views & art view speed, but I only use folder view (rare exceptions) & in that regard Media Server does it better. It also has some more edit/view options but I did not have any need of this. I had both running for quite a while & did flip between them but over time I prefered Media Server. When DSM-6 came along & the added hastle of installing Minimserver & Java I just did not bother.
Setting up and managing a streaming server is no more or less complex than setting up and keeping a turntable in good order. Less in many cases since so little care is required once the server is up and running. After setting up a box as a NAS and Asset/Plex media server I can't have logged into it more than once in 18 months. Its always on and just works. In another decade I might swap out the drives and fans and battery.
It really depends on what you are comfortable with. The OP sees complexity and fiddly maintenance. I see the same thing every time I look at an LP12.
Bear in mind that as time moves on, having a NAS with UPnP capability will be so ubiquitous due to their use beyond streaming music that eventually buying a streamer will incur little or no effort to get music flowing. ... for 'most' people.
Huge posted:garyi posted:I use daphiile music server running as a virtual on windows 2012 running on a dell 2900 server with 32gigs of ram and twin quad core zeons.
I have set the 8 drives up in 4 sets of striped, (0) so its super fast, this all backs up to an proliant server via cobian back up (great, free pc back up software)
Not overlay complicated to set up, I only got into virtuals this year but will never go back now. If I select an album to play on the uniti music starts half a second after hitting play, gapless and to date not so much as a hicup out of place.
In my experience the best servers are powerful servers. Their downside is electric use and noise. Mine lives in a garage so noise is not an issue. I help the environment by trying not to fart too often.
I think most people would find that complicated. A large pot will suffice to hold a small volume of liquid, but it's not the most efficient way of doing it.
Compared to a NAS, then for audio use, your server is a bit like using a 100l high pressure PTFE lined fermentation tank to hold 1/2pt of beer!
(But you may well have other applications that justify just such a server.)
P.S. how do you protect against Ransomware Viruses?
Indeed some would find it complicated. Some find setting up a QNAP complicated too.
For people with a bit of knowledge, i.e comfortable installing an operating system, then a server is just a computer nothing more, nothing less.
My server is just that, a server. It serves up all sorts of stuff, I VPN into it for files, it has PLEX running and dishing up tv shows and films, Daphile handles music. The lad even has his own virtual for running a minecraft server.
Ransome ware viruses? I also run my own router on a dell optiplex, because a PC is very much more powerful than commercially available routers, it has a full suite of tools running on it called UNtangle, which is on the look out for crap like this.
Plus my server is not publicly available, none of the drives are 'mapped' on other PCs and it has no email/browsers running on it. My router runs 100% stealth and my kids PCs are on a different sub set of the network because they are kids!
Whilst I consider myself 'handy' with PCs, I learnt all this stuff quite easily. I have to admit that for the most part windows is easy. linux is hard. But Untangle was a dead easy installer, and virtual servers, enabling me to use one piece of hardware for what would otherwise be five pieces of hardware is amazing tech I would recommend anyone try.
Hi Garyi,
Yes, I also believe having an incremental backup drive that isn't mapped to exposed computers is the final key to defeating Ransomware: I use a non-mapped drive on my NAS for this.
I suspect that (by comparison to most) you're 'gifted' in your ability to 'just see' the patterns of the way computer stuff fits together. Most people can't do this, often they don't have the ability to visualise it.
I suspect that if I had understood the importance of PCs back when I was at school (The Apple Mac Plus had just been released, and windows 3.11) then I might have taken a different career path, however now I see an industry where I couldn't imagine how anyone could get in (thinking of my sons rather than myself)
I started with VAX-11/VMS and then progressed(!) to MS-DOS 2.31.
In terms of getting in to the business now, the best route from the technical angle may be DBA, systems analysis, business analysis, network administration or possibly Tech Support roles then moving on to system design or system planning. The other approach is project management, but that has little to do with computers. The biggest problem with access to code based jobs (leading on to software design areas) is that almost all jobs, even at entry level, require 1-2 years experience (which you can't get now as you can't get an entry level job to get the experience!).
Competition for programming jobs is too extensive from places like Mumbai. However there's a problem there, many are trained in too narrow a range of understanding - we repeatedly got sent "experienced" Unisys assembler / COBOL programmers who a) didn't know that the data word of the machine was 9bit 1's complement, b) didn't even know what that meant. However the agencies still convinced managers (who also don't understand the technology) that these people actually are top class programmers (and we repeatedly had to fix the mess they left behind).
Knowledge Is Dangerous*
Regards
Pete
Let the professionals handle the storage, tagging and scaling thereof - to wit, Tidal, Deezer and (to a lesser quality) Spotify provide the source; all you need is a capably-designed streamer (and a paid-tier account) to enjoy it.
David, that doesn't work for me.
To the best of my knowledge none of the professional organisations fully comply with the Equality Act 2010 (N.B. I wouldn't expect them to) by providing a personal fully customisable choice of selection criteria. Being dyslexic, I don't necessarily classify or identify things in the same way as the majority of people (in fact, very often, I actually don't use the same hierarchy of criteria as any of the other people I know).
So: Not always a viable option.
I believe it's important for a variety of options to be available to all people who need them. I'll keep my server (on a simple NAS).
DavidDever posted:Let the professionals handle the storage, tagging and scaling thereof - to wit, Tidal, Deezer and (to a lesser quality) Spotify provide the source; all you need is a capably-designed streamer (and a paid-tier account) to enjoy it.
Except that for those of us who prefer to own and not rent, and to be isolated from the inevitable ups and downs of online providers and connections, what is needed is a stable music store at home, that (ignoring the technophiles amongst us) is easy to set up by ourselves, or has simple infallible instructions, or is readily set up for the dealer selling the streamer, that is then stable and easy to maintain indefinitely, and of course with no adverse effect on the music replay.
From the constant discussions, and cries for help, on these forums, many supposed storage solutions with different NAS and computer machines, and alternative choices of UPnP server software, and network cables, switches etc, it is clearly far from what someone wants who just wants to take his or her new streaming device home and listen to music. As I've suggested before, Naim should require any dealer selling their streamers to offer a turnkey music store solution, (or more realistically, a small choice thereof).
And I remain wholly convinced that the simpler approach not playing files over a network unless you want to is better still, with rendering software running on the same device as the music store, whether that be, say, an instantly ready Melco, or Mac Mini-Audirvana as I use, which requires a little setup - but not much if you buy the machine with appropriate drives and ram installed rather than DIY as I did.
Garyi........ and my kids PCs are on a different sub set of the network because they are kids!
Mr Underhill posted:Garyi........ and my kids PCs are on a different sub set of the network because they are kids!
And my (now 20-something) can set up and manage servers and the network as well as I can, or if the truth be told the older one can probably do it better than, and certainly quicker than me!
IB, I think the principle of a 'turnkey' network option is a sound one, maybe...
1 x NAS (QNAP, or Synology), option of 1 or 2 bay, option of 2, 4, 6 or 8TB.
1 x Unmanaged Network Switch
1 x Wireless Access Point (with DHCP)
4 x generic 3m CAT 5e cables (option for 10m)
Including...
Setup / installation of the NAS (with 2 items of public domain music), switch, streamer and WAP (with nothing else is connected to it - all other connections are the user's responsibility)
Demo of playing the music using the IR remote.
Demo of the iOS/Android app (provided the user provides a tablet already connected to the internet and the app store and provided the user then connects their tablet to the network via the packaged WAP)
Optional disabling of the WAP's DHCP server if the user wants to use their own broadband router.
I think anything more than this is asking the dealers to get too far into user's custom network installation.
The problem is that if this fails outside of the pre-configured demo (for instance the app cannot use discovery to find the streamer or NAS after connecting the user's broadband router), the dealer could legitimately claim that the user's router or network isn't correctly configured or is interfering and that's the user's responsibility. Tying it down as a turnkey system limit's the dealer's responsibility to the supplied components.
Yes, and the challenges arise when someone already has a part system, such as a WAP and a network of sorts: but the principle should be to be able to provide for someone who has nothing if they want it, and maybe be able to supply a configured NAS for those who have a network.
Simon-in-Suffolk posted:Jason - when I hear a difference and people reporting x sounds better than y - sometimes its chatter, noise and self justification - but other times when you see a correlation there might be something in it. This is what I have attempted to look at and better understand; by removing the clouds of marketing and sales obfuscation and sometimes over simplification and look underneath the covers at some of the engineering and technical parameters of streamed audio.
My initial findings have shown a correlation of certain network data 'dynamics' with SQ. This sort of thing is not necessarily unusual or strange in my professional world - so I have applied some of my thinking of that world to the world of domestic audio replay systems.
So far I have been able to replicate and and analyse to confirm the likely contributory causes of some of the best SQ I have heard. To date some of the best SQ has been achieved by using a very simple modestly powered NAS and combined DLNA media server simply offering Wav PCM files without transcoding from FLAC.
So far I have been able to reasonably well predict a data flow pattern based on the heard SQ. More to do..
Simon
Simon,
Will be interesting to see how you get on and see if there is a general correlation between a simpler network setup and better SQ.
All the best.
Hi Jason, indeed, it's not so much simpler network, but a simpler media server / NAS setup seems to, in my examples, provide a more consistent inter packet timing within bursts of transfers and when the data transfer is of a fast enough rate encourages the Naim streamer to use the TCP ZeroWindow flow control mechanism which seems to get the best SQ out of the Naim streamer.
Simon
Mike-B posted:Jason posted:Something I find interesting is using my Synology NAS, simply switching between Synology's own media server and Minimserver, I find a tangible difference where I prefer the SQ of Minimserver! This is using .wav files stored on it, with no transcoding.
I've tried this a number of times & have not detected any change in SQ that I can conclude as either different or better.
Using my NDX all wired via a switch from Synology DS214 (CPU Dual Core 1.066 GHz & RAM 512 MB DDR3) I've used FLAC straight & transcoded & WAV with both Syn's native Media Server & Minimserver.
Media Server is rubbish with transcoding, but as I'm all WAV & DSD, I tried it for test purposes only.
With WAV the only difference I can detect is Minimserver uses more RAM & CPU than does Media Server.
The upsides with Minimserver are transcoding, plus it does a better job of art views & art view speed, but I only use folder view (rare exceptions) & in that regard Media Server does it better. It also has some more edit/view options but I did not have any need of this. I had both running for quite a while & did flip between them but over time I prefered Media Server. When DSM-6 came along & the added hastle of installing Minimserver & Java I just did not bother.
Mike,
Ive no idea why I can hear a difference between the two. It's not huge, but it has been repeatable and noticeable. I suspect as in most cases, whether this is repeatable in a different system in a different house and with a different network is another thing. I don't really have an answer as to why, but it will be interesting to see how Simon gets on if he uses different servers within the same hardware when he does his testing. Strange game eh!
Simon-in-Suffolk posted:Jason - when I hear a difference and people reporting x sounds better than y - sometimes its chatter, noise and self justification - but other times when you see a correlation there might be something in it. This is what I have attempted to look at and better understand; by removing the clouds of marketing and sales obfuscation and sometimes over simplification and look underneath the covers at some of the engineering and technical parameters of streamed audio.
My initial findings have shown a correlation of certain network data 'dynamics' with SQ. This sort of thing is not necessarily unusual or strange in my professional world - so I have applied some of my thinking of that world to the world of domestic audio replay systems.
So far I have been able to replicate and and analyse to confirm the likely contributory causes of some of the best SQ I have heard. To date some of the best SQ has been achieved by using a very simple modestly powered NAS and combined DLNA media server simply offering Wav PCM files without transcoding from FLAC.
So far I have been able to reasonably well predict a data flow pattern based on the heard SQ. More to do..
Simon
Simon, I seem to recall that you favoured running UPnP server software on a separate device (was it a R Pi?) Are you saying that you have moved on from this to favour running the server on the same NAS you use to store music files?
Simon-in-Suffolk posted:Hi Jason, indeed, it's not so much simpler network, but a simpler media server / NAS setup seems to, in my examples, provide a more consistent inter packet timing within bursts of transfers and when the data transfer is of a fast enough rate encourages the Naim streamer to use the TCP ZeroWindow flow control mechanism which seems to get the best SQ out of the Naim streamer.
Simon
I think I'm with you, in essence a consistent flow making the streamers job easier, OK. Don't want to stray to far from the OP's original point, but would be interesting to see if that 'flow' differs when simply using a different upnp server.