Servers: what do you love about them? What do you hate about them?

I’d like to hear about experiences with servers: what’s worked well for you? What has disappointed/frustrated you? What do you hate about them?

Let’s hear it!

Dedicated made-for-the-purpose music servers?

Building a DIY PC server and then tinkering with it is good for a “nerd-on” and has been a satisfying learning experience. The commercial options are made by smarter people (than myself, anyway) and are supported to varying degrees when you have problems or need updates. You are likely to have a more trouble free experience with them and they are, presumably, made to sound good.

If you use a PC or Mac and ask it to do other things for you then your experience may not be as satisfying and a commercial unit is probably a better idea. Overall I would give the tip to DIY PC servers over Macs for sound quality, but you are going to have to work to get there. The difference is in the ability to manipulate the software environment, including control over the hardware by way of BIOS. If the idea of masochistically dueling with your PC does not ring your bell then either platform will do. I love Mac for everything else in my life.

“Server” could include at least four types of equipment:

  1. a purpose-built piece of audio equipment designed to store and play music files (often running Linux)
  2. a Mac or Windows machine devoted to playing music and tweaked appropriately
  3. a Mac or Windows machine used for general computing as well as audio
  4. a NAS

I own 2) and 4). For a long time I have used my NAS as my main music source and have been very happy with it, both in SQ and functionality. After buying a PS NPC, I needed a computer in the music room to rip vinyl so I bought 2), a small headless Windows machine . It never sounded as good as my NAS for playback until I tweaked it with USB cables and Fidelizer; now it’s close, and I could perhaps improve it further with more work. The NAS, feeding Bridge II, is still more convenient to use as well as marginally better sounding. I agree with wglenn that using a general-purpose PC for audio is probably not the best path sonically.

Type 1) items are often advertised as easy to use – insert a CD and the machine rips the files, gets metadata from the internet, then stores the files ready to play. If you are a serious classical listener, don’t think for a moment that the metadata piece will work for you. You have to create your own tagging system and massage whatever you get from the internet. Even if the machine does find the right album, the data is almost certain to be inconsistent (Beethoven, Ludwig van? Van Beethoven, Ludwig? Ludwig van Beethoven?).

Elk: preferably, though adapted Macs or whatever are of interest, as well. Intent is to ferret out what the common problems are with such things, as well as the best points.

I’ve been using a NAS (Synology DS412+) for a couple of years. I run MinimServer on the NAS and use a Sonore Rendu to convert the ethernet ‘packets’ to SPDIF for input to my DAC, with BubbleUPnP as controller on a Nexus tablet. Very happy with both the sound quality and ease of use, particularly as I only need to use a general purpose computer when ripping or curating the music files.

Setting up the network for connecting the NAS to the renderer can require some work to ensure no drop outs and good isolation of the computer/disc hardware from the hifi. Thus I wired as much of the ethernet as possible, switched off all but the essential wireless devices, and moved the latter a couple of metres away from the hifi. I inserted optical isolation into the last wired section of the ethernet connecting to the renderer. No great expense required here: just a couple of ethernet cables, a dual fibre optic cable and a pair of TP Link wire-to-fibre converters (connected back-to-back by the fibre optic cable). Easy to set up and a definite improvement in sound quality.

davidl, I have a 412+ too, could you elaborate a bit on the fiber isolation gear, where I could source this, ect, Thks.

‘I inserted optical isolation into the last wired section of the ethernet connecting to the renderer. No great expense required here: just a couple of ethernet cables, a dual fibre optic cable and a pair of TP Link wire-to-fibre converters (connected back-to-back by the fibre optic cable).’


Got the idea of using fibre-optic isolation from reading a blog by Andrew Everard. Details are here:

I managed to source some inexpensive linear power supplies for the TP Link converters in the UK - I guess you should have no problem getting these in the USA.

Hope this helps


Just ordered the converters from Amazon and the cable from Cables For Less. Thanks, davidl. Can’t wait to try this out!

wglenn said Just ordered the converters from Amazon and the cable from Cables For Less. Thanks, davidl. Can't wait to try this out!
Pls let us know your impressions. Wondering how this would compare to the sonore LAN Noise Filter, I have neither.

Sure will, but I don’t have a LAN filter either. The boxes are here but the fiber optics probably won’t arrive before I head for San Diego but I’ll try them as soon as I get back.

Arghh. Again, I say, aarghhh!

So, in my haste to order prior to going out of town I did not pay attention to the “Gigabit” designation of the transceivers and it appears that they will not function with the Bridge(II). I suppose that I could throw a network switch between the little box and the Bridge but that seems to be a little self defeating. Any ideas, or Amazon return?

EDIT: Yuh-OK, I ordered the 10/100 versions. I’ll let you know…

Spent many years as a full-time server guy. Made the right choice, overall, in switching careers, but I do miss it!

For the moment though and en pointe, I wanted to offer a link for those who are using Windows 10 in your music servers and are, like me, annoyed by the inability to keep it from rebooting itself whenever it wants to to install updates, or after your period of time to schedule a post-update reboot runs out. I find this feature to be a real PITA, because I am often away, don’t want to have to remember to remotely log in to check my server(s) from time to time to see if a reboot needs to be scheduled, and if the box reboots in mid-day it will induce an acute episode of wife-server-panic-disorder syndrome and I will get an unhappy phone call wherever on earth I am. if they simply let you pick a specific time each day for any reboots to happen, that would be fine, but that would be far too straightforward. This link has some ways to deal with it:

I don’t believe that any of these things will prevent security updates from installing, and they will not work on Win 10 Home. AFAIK (some of you may know better) the only way to completely prevent updates from downloading/installing if you have an ethernet connection is to disable the windows update service. But if you do that you’ve gotta remember to fire it up every so often to check for new updates, so I hesitate to suggest that. You can mark your internet connection as metered and prevent updates that way, but only for wifi connections, not ethernet.

EDIT: Should be easy enough to write a powershell script to disable and enable the windows update service, which you could then schedule to only be enabled at specified times.

Got my (10/100) converters and hooked them up which was easy with no fiddling required. I am running mine between a network switch and the DS DAC which is how I connect to my PC server. The server, network switch and the server - side TP Link are on a different AC circuit than the DS DAC and it’s TP Link and I am using the TP Link switching power supplies on both ends. My setup differs from davidl’s considerably in that his optical boxes come prior to his Rendu, which is then connected to his dac by SPDIF, so results may vary considerably. It was a little difficult to a/b compare the configurations because each switch required a bit of time for the network to readjust (mConnect wireless in particular) but in my case it did not matter too much because I only went back and forth about three times.

The difference was very noticeable, with my preference being for the direct ethernet link in my system. Narrowing and loss of soundstage depth with the TP Link was the most remarkable change, accompanied by a slight “etching” of instruments that made them stand out more individually but ultimately less likable to my ears. With my system configuration, the TP Link did not produce acceptable results. What I do like, however, is the thin optical cable. In my loft, where my stereo is, there is no ethernet connectivity so I have run a cat 5 cable up the cold air return which then runs across the carpet to my network switch. My wife loves thisshaking-head-no-smiley-emoticon_gif and will probably miss the big black cable when I replace it with the much thinner optical one to provide a link to the router for wireless control of the server (I did not want wireless built in to my server for noise reasons).

Thanks for the detailed description of your experience with the TP Link wire-to-optical ethernet converters.

A couple of points.

Firstly, I’m not clear how your connection topology differs from the one I use in the region of your DAC. How do you connect the second TP Link to the DS DAC? Do you use wired ethernet direct to the Bridge II in the DS DAC?

My complete connectivity is:

Server (wire)> switch (wire)> airport extreme (Wireless bridge between floors)> airport extreme (wire)> switch (wire)> TP Link(1) (optical fibre)> TP Link(2) (wire)> Rendu (SPDIF)> DAC.

Secondly, in my original description of the setup I use I forgot to mention that I replaced the switching-mode power supplies that came with the TP Link converters with a pair of linear power supplies (as Andrew Everard mentions in the blog I referenced above). I’ve not done a comparison of the sound quality of the setup with switching-mode and linear power supplies but I’d guess that the switching-mode power supplies could account for the degradation you heard.

My setup is Apple Airport Extreme for the wireless control point>Network switch. The switch is also connected to my PC server and DS DAC. Sounds like the difference is that there is an intermediary device between your network and the dac, the Rendu. You could consider the Bridge to be an intermediary device as well I suppose. Technically it is, but being integrated into the chassis of the dac I tend to think of it as part of the dac. In the end, they are two different devices and likely transport the signal differently. The communication pathway is certainly different, SPDIF vs (?) I2S. Just guessing that the transfer of data internally between bridge and DS Dac is by I2S. Yes, the final connection is by ethernet cable whether it spans between the switch and the Bridge or between the second TP Link and the Bridge.