Need help moving from CDP to Mac Mini music server


I've decided to move try a music server solution because my listening habits have changed (maybe the ipod lifestyle has ruined me). So the time has come to sell my beloved Sony SCD-1 and I need some help figuring out how to implement a sonically comparable (or better) music server solution. My system is an Aesthetix Janus with a BAT VK-250 feeding a pair of Totem Forests. I'm currently thinking of getting a Mac Mini with a 500 GB HD and streaming ALAC to my airport express, then using the toslink out to feed a D/A (likley a Benchmark DAC-1) which would run balanced to my Aesthetix. I'd then use my powerbook to control the mac mini through bonjour (although I can also use my plasma as a display).

My first question is, can anyone familiar with using the mac mini server solution comment on the sound quality, especially in comparison to a quality CDP such as the Sony. I want the convenience of having my whole library be accessible, but I want to definitely don't want to sacrifice the sound quality to do it.

Secondly, I've read that the digital out on the mac mini has higher jitter and that using the usb to output digitial is better. Is it also preferable to utilize the usb method instead of using the Airport Express's optical out?

Thirdly, does anyone recommend a specific external hard drive to use with the mac mini? There seem to be a bunch out there.

Lastly, how do people feel about the Benchmark DAC-1 in this setup? I want to keep the whole budget under 3000 w/computer, external drive, DAC and wires, so any suggestions on DAC would be welcome (heard Stello makes one with a digital input).

Thank you so much in advance for your help,

Matt
mimberman
Matt, please post the exact link to the unit you are talking about, I couldn't find it on the site. The word STRIPE in your description has me worried. Striping offers no data protection, it's strictly for increased performance, which is not necessary for streaming audio. As for the confusion concerning the Terrastation R5, the confusion is coming from the two different protection mechanisms. RAID 5 gives you usable space on roughly all the disks minus 1 (parity disk). The implementation is rather confusing, see Edesilva's explanation above. Mirroring you get exactly 1/2 of the total amount of available space for use, one disk has an exact duplicate of the other. Both offer data protection against loss of a single disk, just through different implementations. The main difference to keep in mind is with R5 you need 3 or realistcally 4 disks minimum. R1 (mirroring) only requires two.

Again, please post the link to the exact device. My concern is if it's listed as 800/400. My guess is they are giving you two 400 gb disks and you can either mirror them via R1 and have 400 gb usable or stripe them via RAID 0 and have 800 GB usable. Striping you write across both disks, but different data. If you lose either disk you lose everything.
Matt, be careful, RAID striping gets you transfer speed, not redundancy. Let me try RAID 5 again. Say you want to store 3 numbers -- 012, 147, and 532. With the four disks, RAID 5 puts 012 on Disk A in Block/Sector X. In that same Block/Sector on Disk B, it puts 147. In that same Block/Sector on Disk C, it puts 532. In that same Block/Sector on Disk D, it puts 691, which is the sum of what is on A, B and C in that location. So...

If Disk A goes bad, it can recover the data for Block/Sector X on that disk by taking 691 and subtracting what is on that Block/Sector on Disks B and C: 691-(147+532)= 012.

If Disk B goes bad, it can recover the data for Block/Sector X on that disk by taking 691 and subtracting what is on that Block/Sector on Disks A and C: 691-(012+532)= 147.

If Disk C goes bad, it can recover the data for Block/Sector X on that disk by taking 691 and subtracting what is on that Block/Sector on Disks A and B: 691-(012+147)= 532.

Disk D, obviously, can be recreated from A, B and C. This is a simplistic version, but RAID 5 is higher efficiency than mirroring because the effective storage you get is (N-1)*S, where N=number of identical disks and S=size of disks. For mirroring, obviously, the efficiency is N*S/2.

Maybe the tera isn't pretty, but mine's in my closet, do I don't worry about it. ;)

Good luck--as far as Mac Mini specs go, I bought a 1.4GHz one with the bigger memory package and it was overkill for doing audio.
Interesting article from 6 moons... in the end it's not a review rather more along the lines of a technical explanation of usb...
http://www.6moons.com/audioreviews/wavelength2/brick.html
You would be much better off using a USB DAC rather than a DAC that is fed through a Toslink connection. Using the toslink and the Benchmark there is always the issue of unwanted variations in the frequency of the incoming word-clock signal, which the converter is always trying to track. But with a USB controller generating its own word-clock signal, the frequency remains steady.
"But with a USB controller generating its own word-clock signal, the frequency remains steady." Well, only as steady as the clock in the USB interface. It is not a panacea, because some USB chips are pretty bad jitter-wise.