The very current state of music servers (I just wrote this article last week, posting here first):
The technology behind Music servers is still brewing and percolating. I would say they are past puberty now, bounding towards young adulthood but still with room to grow and time to physically and intellectually mature. They are all built from the same parts: A hard drive storing the music as an Ethernet based NAS or as an internal hard drive (spinning or solid state), a computer needed to run software that allows selection of the music to be listened to, some type of translation device to convert that selected music data to a language and conduit the DAC can talk to and understand (either directly outputting USB or requiring a translator called an Ethernet Renderer that essentially converts those little magic packets of Ethernet information into a more standardized USB signal), some form of distribution hub, and a power supply. As with all audiophile gear, each component in the system including the cables that connect those components, the quality and speed of the parts used, the internal and external shielding from noise those components (and cables) are sensitive to, and the quality of power supplied to the equipment all determine the final performance of the system. Two solutions seam to be most common, and two methods of data transmission; although at this time one of the two methods must, at some point, be converted to the other before being able to speak to the DAC.
Music servers now exist as either independent and purpose built components with everything in one box (or one box with an external power supply) or as essentially separate components with everything above separated. The same task is accomplished either way. In the end, both must output either spdif, AES/EBU or, most commonly, USB. At this time most DACS don't accept Ethernet as a standard input; and those that do still require an Ethernet Renderer inside the DAC to convert the Ethernet packets into USB or bypass USB and convert directly to I2S which the DAC chip can understand.
An all in one music server internally houses and utilizes either Microsoft, Linux or a Mac operating system (usually Microsoft or Linux) and many have opinions about which is the best. Separate systems either use the simple Linux operating system built into the NAS or actually use a computer running Linux, Windows or MacOs that has been modified to use only those computational systems that are required to run the software which allows you to select the music. These computers are customized or custom made. A similar system has been designed and built into a dedicated music server. The separates system utilizes Ethernet to communicate with all of its parts until the final step. Ethernet is known to have galvanic isolation advantages and noise isolation advantages as a result of how packets of information are sent and reconstructed, which demands that all the information first arrive and then be converted. This, theoretically, results in less data transmission errors in a system isolated from certain types of noises. The problem is that each component along the way is still sensitive to noise and sensitive to the quality of its power supply. A dedicated music server houses everything in a single, shielded enclosure and shares a power supply, or several power supplies. Dedicated servers output their signal to the DAC via either spdif, AES/EBU or USB. USB sends its signal in a more linear fashion and is more susceptible to data errors, and is not galvanicly isolated. It does, however, have its advantages and many who are in the know believe that USB is still a better signal path and has many years of usefulness left before it goes the way of Betamax and the cassette.
After the computer and software point the music in the hard drive to the converter it needs to be beaten, manipulated and "reclocked" to essentially clean the data stream and get the most perfect set of bits possible to the final step. Again, this can be done with a separate device or internalized within the single chassis of a dedicated server. It then must end its long path by being converted via what's called an Ethernet Renderer from its little magic packets into standard USB to be passed on to the DAC for final conversion.
A dedicated server transits all of this information within. Some of it is, in fact, transmitted via Ethernet but many convert to USB or direct to I2S as quickly as technologically possible. Long Ethernet runs don't degrade sound at all (which is a luxury of a separate system since components can be placed far from the listening area), but very short runs of most other signal type yield better (i.e. less noise) signal transfer. Most prefer solid state media storage these days, whether internal or external. And most prefer their operating software on a separate solid state drive from the storage drive. There ends the descriptions and pitfalls of a dedicated Music Server. A system of separate devices is reliant on one more device, the Switch which allows everything to talk to each other in super fast two way data lingo that listens to the software running on the computer to direct those bits of music data on your hard drive to run as fast as they can to your DAC and get bounced around, attacked by noise, reclocked, and eventually converted to USB so your DAC can take that final step and make good old fashioned analog music signals with them. The Switch can utilize either copper wire or fiber optic wire to transit the Ethernet signal (optical obviously adds an extra layer of noise isolation but also adds more boxes to muck with the data stream). All of these boxes then require high quality linear power supplies to make the entire system sound its best. The NAS, computer, Switch, Ethernet Renderer, Reclocking devices and any optical converters used ALL require moderately expensive and well built power supplies. And they ALL require high quality (i.e. costly ) cables to connect them. As a result, a simple separate system including a NAS with built in Linux running music server software can send its signal to a Switch (which every house these days has anyway) and then to an Ethernet Renderer for USB conversion to a DAC, and there you have a relatively cheap and very good music server. To put together something more reference level requires top notch power supplies (Paul Hynes seams to be the resident expert and supply source), customized computers with solid state drives, lots of Reclocking, a good switch and an Sotm or MicroRendu Ethernet Renderer with lots of good cabling throughout. I believe the final separates system's result is WAY cheaper then the average top tier single box purpose built Music Server but requires quite a bit more effort to select, assemble and implement. Can a system of separates sound better then a purpose built dedicated Music Server? I don't see why not. To a tweeker and tinkerer it would not even be considered effort to put a high end multi-component system together. To someone on a serious budget it would be a worth while effort that would be paid back with huge dividends. To, I believe, most audiophiles, its not worth the complexity and DIY feel that putting together a true high end Server system would require. I could be wrong though, considering the audiophiles propensity towards "separates" like amp and preamp, or DAC and transport.
As a small aside, and to be through, someone has come up with a way to run the computer/software box directly to the Ethernet Renderer by utilizing 2 Ethernet ports and bridging them. The results are reportedly a significant improvement in sound quality by bypassing the need for a Switch in the signal path. I'm not sure how complicated this is to achieve, but the concept is logical and removes a box that touches the data, can add to its corruption and offers another point for noise to enter the system. I can refer anyone interested to the on line thread this idea is discussed.
Dedicated servers have now been out for years and those that make them have revised and tweaked these hardware and software packages to high levels. As I said, they are no longer anywhere near infancy. Some have come out with packages that are, sort of, half way systems with Servers that house the computer, software and typically a storage solution and export Ethernet to be rendered externally as more DAC's start to become available with built in renderers and more (and better) external renderers become available like the Microrendu and soon to be released SoTm SMS-200 ultra. But a fully equipped dedicated Music Server can now serve up the goods in true Audiophile fashion. And if you do have the know how, financial restrictions, DIY desire or tinkering prowess to put together a reference level multi-box Music Server then you will be equally (and some think more then equally) pleased with your results.
