Like many of you, I was ripping and encoding Mp3's many years ago.
And although I never used Xing, I can't say my 128 bit Blade rips are making me happy today.

I have come to a crossroads, and am not sure which path to take. On one hand, I am really happy with how LAME -aps sounds, and of course like the size. (Of course, I remember saying the same thing 5 years ago. Will I say it again in 5?)

On the other hand, I can go the lossless route and be assured of a great archive.

My problem is that I have just over 10,000 cd's to convert. The storage requirements in going with something like FLAC is quite large. I estimate only getting about 380 albums on a single 120GB drive (around $90). DVDR media actually comes close to the same cost. And CDR format was already out of hand for MP3 size (1200 CDR's stacked up).

To me, I feel like the only option is to wait until 1TB drives cost $99. (Not going to happen) or got with LAME -Aps.

10000 cd's .. wow..

aps seems to stack up pretty well, no complaints from me, wonder how much space it will take for that kind of data @ ~ 200 kbps .. prolly more that 120 gig, much more

MPC is a great middle ground for you that has quality arguably better than mp3, but with comparable sizes. You might want to give that a try.

Just as a reference, you have over 10,000 CDs. What are you, a DJ? That's over $100,000 spent on CDs. If you drop another $1000 or so, you could easily get a 1TB Raid-5 array for yourself.

Also, as another reference, assuming they're all full-length music CDs, you have about a year's worth of music sitting in CD form.


Sorry, didn't mean to be offensive, but that's a LOT of CDs. I have a hard time visualizing 10,000 of something like that.

By the time you'll be halfway through ripping, chances are the available HD space is tripled and the cost is a third of what it is today!

My solution is letting Musepack shrink the vast bulk of the CD's that pass me by, and keep the most important ones (for any reason) losslessly compressed with FLAC. Every time I reevaluate my decision (which is sometimes too often) I end up being quite happy about it.

Cheers,
Uosdwis

I only have about 2500 CDs and I'm already in such a conundrum, but I'm not as bad off as you, of course I just need to pick up a 250GB to hold all my LAME rips (aps) plus plenty of movie rips.

50minutes/album avg. * 60 seconds/min * 200kilobit/sec /8 byte/bit = ~75MByte/album
2500 albums at around 75MB per is roughly 187Gigs, and I have 80 non-OS, which is stuffed with the rips I haven't burned yet

You're looking at closer to three-quarters of a Terabyte with aps rips

If archiving is your primary goal with ripping, and you're largely going to be playing the rips from your computer, it seems like MusePack/MPC would be the best option for quality and size.

The ideal option is probably MPC on a large RAID 1 array just in case one of the Hard drives takes a dive. Which happened to my first 30Gigs of rips
Edit because I dont rip movies with alt-presets and because RAID 0 is not redundant

To think I was getting rather proud of my 300 cd collection... sigh...

at least I can say they are all really good discs!

yeah, I got interested in FLAC. Did a quick test. Some of my old Rush CD's came out around 250MB. OK, not so bad. That might work.

Vapor Trails came out around 500MB. I figure the middle ground would be around 350MB. That just won't work when you're talking about archiving 1000 (or more) CD's.

Thanks for the nice comments, everyone. That's why this site is the best!

sthayashi, as Audible mentioned, a TB worth of storage will be needed just to house that much as simply MP3-aps. That is certainly affordable for me (although pushing it), although I'd probably wait until the next size up from 120GB breaks the $99 mark, and with Serial-ATA. But contemplating the cost of hard drives to hold the Lossless -- no way.

I think the only reason I was even thinking about lossless archiving as a solution was because when I did the math, it looked like 8 or 9 FLAC albums would fit on a DVDR. This of course is close to the ratio of mp3 albums fitting on a CDR. Which back in the day, I didn't have any complaints about! (I don't know if I'm alone in this, but discovering mp3's back in 1998 was kind of like having a childhood Christmas, a Birthday, and sex for the first time, all rolled into one!)

The 1200 CDR's I have stacked up in spindles doesn't bother me even now, so would 1200 DVDR's bother me? Not if the price was the same as CDR's are. That'll take a while, but can I get 4X DVDR for $1 each now? That seems affordable at staggered intervals (which luckily, this hobby takes a while to do)

I've recently started my education on MPC, so will have to ponder that.
My first mistake was looking at the main web site of it (all black background, millions of useless links)
Since computer playback is the main use here, MPC sounds like a sound fit. However, the popularity of MPC may hamper my style. Not that...*cough*....I'd ever trade these... *cough*... But if some other Slider's World I did, MPC might not be any good.

Concerning the cost of hard disk capacity, I thought I'd add something...

I bought my first *real* PC in college (not counting a Commodore VIC20 as a *real* PC, but it was fun!) It was 1988, and the machine was one I bought the pieces for and assembled myself, component-by-component.

Here are the specs:

Intel 286 / 16 MHz
1 MB RAM --- 640 kB was pretty standard at the time, but I sprung for the *big* memory option
40 MB HDD
5 1/4" floppy --- giant disks which stored 320 kB to 640 kB

...well, you get the picture.

Anyway, that 40 MB hard disk cost me about $300!!! That comes out to $7.50 per MB.

-----------------------------------------------------------------------------------------------

Fast-forward to 2003. Just for a comparison of new technology, my portable MP3 player takes Memory Sticks (flash memory) which are smaller than a stick of gum and hold 32 MB up to 1 GB. The biggest I own are 128MB, which now run about $50 each (~$0.40 per MB).

As you stated, a 120 GB hard drive now costs about $90. That's $0.00075 (75/1000 of one cent) per MB.

So we've seen a price drop of some 1,000,000%. A ONE MILLION PERCENT price drop per MB of HD storage in 15 years...or an average of 66,667% per year.

Now, price fluctuations aren't consistant year-to-year, so if you made a realistic graph, the the curve would be pretty bumpy. Hence, you couldn't apply the same formula to a small time frame (<5 years)...so instead...

-----------------------------------------------------------------------------------------------

Fast forward to 2018...fifteen years in the future.

If the same formula applies over then next fifteen years, then HD storage space will cost 1/1,000,000% of what it does today. That comes out to $0.000000075 per MB, or $0.000075 per GB...or $0.075 (seven-and-one-half cents) per TeraByte...or $75 per 1000TB. (OK, so what do you call 1000TB??? A MegaGoogleByte? Actually I think it's a "QuadraByte", but I'm just guessing.)

-----------------------------------------------------------------------------------------------

Anyway, my point is that you can count on the cost of HD storage dropping DRAMATICALLY in the coming years, with the biggest price drops hitting an existing technology right when a new technology hits the market.

So whether that *next* technology will be holographic memory, or quantum laser static buffers, or Star Trek antimatter matrices...the effect will be the same, those spinning disks we call "hard drives" will become as cheap as dirt.

Just something to think about when you consider audio compression technologies going forward.

-----------------------------------------------------------------------------------------------

And BTW...that's a H*U*G*E collection you have...puts my ~350 CDs to shame

10000 cds?
AVG ~45 minutes and 800bits/sec would mean 2.7TB which would need 12 250GB drives

Stay away from 120GB drives, might save a little with $/GB for the drives but you'll need more space, raid controllers, and just wiring is gonna be a mess .

250GB SATA drives, on 2 raid cards able to handle 8 drives.

You could just get 16 drives at once and leave alot of room for expansion, it'd be a fixed cost just gotta go all out at once hehe.

I'd wait a litlte longer till 250GB sata drives drop below $200 (you can find pata ones under 200 but dunno bout sata) and also check out sata raid cards, havn't looked into it much. SATA supports hotswap and all and if a drive ever failed it'd be nice to have a card that supports it and would make it easy to rebuild the array.


Using MPC would save you alot of space / money but in it also costs you your time, if you ever rerip them all again to lossless in the future, cost of your time would pay to do it now


---edit---

"5 1/4" floppy --- giant disks which stored 320 kB to 640 kB "
hehe I dunno if its funny or sad that a 5.25 floppy needs an to be explained

Its gonna be quicker encoding 10,000 cds as MPC instead of lame aps. 10,000 is ALOT of cds, you gonna rip these one by one? that would take forever !!

10.000!?!?!?! didn't you mean 1000 ... 10.000 CDs is about 200.000 dollars on CDs only!

I've enjoyed reading this thread since I'm going through something similar but not on the same scale.

I just bought 4x120GB UDMA drives on sale for ~$80USD ea and built a cheap server using RAID 0+1 & a TB Santa Cruz (also on sale for $40) specifically so I could re-rip ~600 CDs. I'm about 200 CDs into it.

I mulled over this question for some time, lossless vs MPC or Ogg and finally decided on MP3 lame 3.90.3 -aps. While the others may be technically superior MP3 is good enough for my ears and I can play MP3s on HW devices such as my RioVolts, DVD player, and other players I'll prolly buy in the future.

What I underestimated was the time it takes to rip lotsa CDs if you want to be as accurate as possible w/ the tagging info. I'm a bit anal about that.

I haven't done it yet, but I'll prolly MP3Gain the whole collection when done. I'm hoping by the time I'm done SnelG will have included support for the more widely HW supported ID3V2 tag in MP3Gain vs the technically superior APE tag - but that's another topic (No, I don't do silly things w/ ID3V2 like include album art - only the fields EAC fills in for me.)

Happy ripping...
Walkman

Very good point. FLAC is also fast, under 10 minutes per CD to rip and encode on my Celeron 1.3 music server using EAC + FLAC.

Nope, at $1/GB you can FLAC 1,000 CDs for about $350 USD plus either a spindle of DVD-Rs or another $350 for backup drives.

I went the FLAC route (750 CDs so far, another 400 to go) so that I'd never have to rip again, and so that I could transcode to other formats as needed without taking a second drop in quality from lossy-to-lossy transcoding.

That's very nice and all, but are you making dupes of your burns, using (add'l) error correction, or just assuming that the burns will last forever?

My experiences with data archiving on CD-Rs have been spotty at best, even with only Taiyo Uden or Ricoh Platinum media, let alone some of the crap I've bought at electronics and computer stores.

In fact, I'd go so far as to say that a dedicated RAID 1 array will probably result in significantly less data loss over 5 years. And, you don't have to spend the time burning a thousand or more CD's only to see some fail....

Again, I'm one thinking of this on a smaller scale. I've had enough CD-R's corrupt themselves on me to not even think of using DVD-R.

I think raid1 or 5 is the solution, as one drive failing doesn't mean any data is lost, so it saves time and effort

my 150cds = 90gb = not a problem

10000cds = 6TB! (or 3.5TB as FLAC)

And the music industry sais people aren't buying enough cd's

Unfortunately for parity-based RAID schemes (like RAID 5), I believe you typically want to invest in an uninterruptible power supply (UPS) if you value your data. Other than that, RAID 5 would be absolutely ideal, assuming you can afford the drives and buy a decent controller.

Mirroring (RAID 1) doesn't have this problem of course.

Why would Raid5 need UPS? I thought you could recover your data with 1 drive dead as with raid1, never saw anything about power issues? :S

Go with western digitals special edition drives with a 3 year warrenty. SCSI drives have 5 year warrentys.

Here are some links that may be useful. I got them off Slickdeals.net. They are Western Digital 200GB for $115 or if you prefer Maxtor 200GB for $130. These prices are after large rebates, but they are the best prices for high capacity drives that I have ever seen. Both these drives are high performance 7200 rpm drives with 8mb cache. I know Western Digital provides 3 year warranties for their 8mb drives. Not sure about Maxtor though. The Maxtor drives are quieter than the Western Digital drives once you enable acoustic management.
Anish


Edit: To get the $115 on the Western Digital, see the SlickDeals website. With minimum effort, it is $270 - $120 mail-in rebate. The Maxtor is $300 - $50 instant rebate - $120 mail-in rebate.

Edit: OK...I need help using the "Quote" feature...just ignore the quote above....that's not what I meant to grab.

HERE'S what I meant to start with...


It's just amazing how fast the technology's moving, that's all. I've met people recently who didn't know what a 5 1/4" floopy was. In fact...3 1/2's are almost going away...when was the last time you put a floppy disk of any kind in your PC?



I'm heading in that direction, too, I think. When I first came to HA several days ago, I was of the state of mind that "FLAC or 320kbps MP3 are the only (compatible) compressed formats for me. *I* can hear the difference with the lower stuff!" But now, just a few days later, I' thinking that anything above --aps is placebo to my ears. Gotta hand it to Dibrom for tweaking an MP3 encoder to the point where it's most basic setting is transparent for so many people.

And compatibility is what's keeping me away from other formats for now. Until my portable device and the DMS in my car can do MPC or any lossy codec other than MP3, then all I can consider are the formats that I have players for (other than my PC).

To keep you going over the next few years:

• 1,000TB = 1PB = Petabyte.
  
  
  
• 1,000PB = 1EB = Exabyte.
  
  
  
• 1,000EB = 1ZB = Zettabyte.
  
  
  
• 1,000ZB = 1YB = Yottabyte.
  
  
  

Although of course when most people talk in these units they really mean TiB, PiB etc, meaning it's 1024.

Well, maybe. Depends on how much further areal densities can be pushed reliably and cheaply, and how long suitable replacements will take to develop. We're seeing a nice drop at the moment because manufacturers are moving to 80GB platters, but we're also seeing access times going up as heads need to seek more precisely, and yields dropping as smaller and smaller defects become problematic. Better keep our fingers crossed

And my, er, ~5 CDs to shame. *sob*

[quote=DaveSimmons,Jul 4 2003, 06:30 PM] [/QUOTE] Nope, at $1/GB you can FLAC 1,000 CDs for about $350 USD plus either a spindle of DVD-Rs or another $350 for backup drives.

[/quote]
Exactly. That doesn't work for me.

No kidding, eh? As a 19-year old, I think I've done okay in having ~100 CDs. Then, to hear someone come on here and have my collection size *SQUARED*(!!) just makes me feel small.

No, my FLACs are on 2 Maxtor 250 GB drives, and I have a couple of WD drives to use for backing them up (with the WDs kept separately from the server).

I figure I can keep copying drives with a very low chance of 2 drives (primary + backup) failing at the same time. Current cost for this scheme ("manual RAID 1"?) is about $0.60 USD per disc, but within 2 years it will probably drop to 0.30, then 0.15, etc.

Comparing to real RAID-1:
Pro - Simpler to setup and maintain, use motherboard IDE or cheap non-RAID card
Pro - backup drives are not in PC, safer from theft, power problems, infection, or other catastrophic server failures.
Con - must remember to do manual backups. I use my 120 GB OS drive as a temporary backup before copying to the real backup drive.
Con - must periodically test the backup drives to make sure none have failed.

$0.60 per CD is reasonable to me to have all of my CDs online (a super CD changer), in true CD quality, and both transcodable and "clonable" as needed.

You should consider using replaygain instead of mp3gain.

MP3Gain is the MP3 implementation of Replaygain. So basically, it is Replaygain.

Back on topic: I have no idea how you plan on ripping 10,000 CD's. First of all, if I bought 1 CD an hour for 8 hours a day, 5 days a week, it would take me over five years to reach a collection that size. Have you listened to all that music, or are you just a packrat?

If you rip the CD's one at a time, assuming you can get a CD done in 10 minutes, that's another 3/4ths of a year of constantly switching CD's during the 8-hour work day, and probably much longer, unless you plan to sit attentively in front of the drive the whole time. If you're serious about archiving this music, you'll need a better solution, such as a CD tower and a cluster of machines to do the encoding. That already costs $5000, probably.

And after investing all that time in encoding, you'll want a good way to back it up. Since time = money, (let's say you're worth $30 an hour, for argument's sake), that means you've spent $50,000 of your time if you copied the CD's one by one, or $6250 if you used an 8-CD ripping tower. Hell, even if you were worth minimum wage, it would cost you less to buy a ripping tower and encoding cluster than to do the job one-at-a-time. Anyway, the time investment justifies a real storage and backup solution.

If FLAC gives 350MB per album and MP3 gives 75MB, that means the initial storage costs for hard drives are either 3,500GB or 750GB. With a nice round number of $1/GB, you save about $3000 by going with MP3, but you lose the ability to transcode or do any real signal processing on the audio, you don't get guaranteed transparency, and you don't get gapless playback. However, you can run 750GB off a standard IDE card, whereas a 3.5TB array will need some dedicated hardware, probably a RAID 5 with multiple redundant hot-swappable drives, just because you'll have drives dying on a semi-regular basis. Still, with the amount of money you already spent on the task, it seems silly to save a few grand and give up so much sound quality and flexibility.

Then, you will also want a good backup system for this. RAID 1 is out of the question, since as DaveSimmons pointed out, you're still vulnerable to natural disasters, power surges, theft, and viruses. You'll need off-site backup. Since your data won't be changing that quickly, you don't need a light-speed backup solution, but you will probably need a whole lot of tapes on a tape drive that costs several thousand dollars. There have been some discussions on 2cpu.com about backup solutions for large arrays of data, but I'm sure there have been better discussions elsewhere as well. Anyway, you'd need a huge pile of tapes, a fast, reliable tape drive, an auto-loader, and a company that will store, test, and protect the tapes. Now, the cost of using FLAC is much greater than using MP3, but I'd still recommend it because locking yourself into a single lossy format limits you so much.

Archiving 10,000 CD's is a huge undertaking that probably isn't worth it for you, especially since you've mentioned your reluctance to spend money (which makes me wonder how you collected so many CD's in the first place). You can't sell the music online because you don't have a license to do so, and 10-to-1 odds says the RIAA will have you in court by the end of the week if you try to share all the albums on a file-trading network. If you're DJ'ing, you don't want MP3 because you can't do any karaoke or extreme equalization efffects to it without making it sound like crap. That leaves you with personal listening (and public performance, if use use FLAC) as your only practical uses for your music. In that case, you might as well save the money spent on ripping all the music, buy an excellent amp and speakers, and just put the CD in a CD player. With all the time you saved by not archiving all your CD's, you'll have plenty of opportunities to sit back, relax, and listen.

I'm sure that instead of buying one CD at a time, he did a MUCH more extreme version of what I do (and what many people do). I budget myself to about 10 CDs per month, and generally buy them all in one trip. This guy probably buys 100 a month (or more, especially if he's a DJ or VERY serious music collector).

As for being a packrat, I agree that that's an *exorbinant* number of CDs, but then again, everybody has different priorities and different interests. Maybe he has 10000 CDs for the same reason I have 8 hand-signed photos of Sarah Michelle Gellar and 5 of Xenia Seeburg. If you love something, be it memorabilia or music, sometimes you just can't get enough.

Or then again...maybe he just typed one "0" too many in his original post... B)

And the work estimates make a valid point...you'd never have the time to archive nearly that many CDs without a *lot* of help and a *lot* of equipment. Hopefully, those are things he's already arranged for...

I don't get why most people insist on sitting still and watching the progress meter while ripping each and every cd? You stick it in - go away and do some work or whatever, come back and check the logs, and then stick in the next!

It cuts maybe 5 minutes out of every hour, if you have a redudant PC at work with you, which many techie jobbed people do, then you can rip 25 cd's a day or something.. it'd take a long time but won't cost $50'000 of your time

I think I'm actually confusing centrally cached parity (RAID 7) with typical one-dimensional parity (RAID 5), so you might be right. If the ECC info. isn't cached in volatile memory during a power outage, it should be okay I think.
RAID 5 should survive a single drive failure very well, but will be a bigger PIA to rebuild than a mirroring (1) array.
Given that we're talking about RAID for data integrity and not performance, a RAID 1 (or 0+1) array should be the cheapest and easiest to implement, build, and fix if a drive dies.

It seems strange to me that you'd propose a RAID 5 array with redundant backup swap-outs and then exclude such a possibility with RAID 1, where it would be faster and probably cheaper (RAID 1 controller = cheap) to do.

A 3.5TB array is very cost-prohibitive, and it's an open question whether the poster has the cash necessary to maintain such a system, let alone implement it. It's also questionable whether lossless is called for in this situation.
It's certainly the ideal case, but the cost differential is enormous here, making the hardware cost rise dramatically (Maxtor 7200 rpm 250GB drives = $265). Possession of many thousands of CD's does not directly imply tremendous current financial means - I acquired my collection over about 18 years and currently drive a used car

This thread has been helpful for me, as I too am about to embark on a similar project (although on a MUCH smaller scale). In fact, I'm facing about 1/10 the effort that Audible! is going thru right now- I've got around 250 CDs I'm planning on ripping. Of course, I expect my collection to expand over time, and therefore am cognizant of starting with a storage solution that gives me room to grow. Which brings me to my question...

I've got a entry-level hi-fi system that I ultimately plan on playing my music on. (Hale Design Group Rev3 speakers, Bryston B60 integrated amp). For quality and obsolescence reasons, I've decided to rip and store all my music as WAV files (no desire to EVER rip my collection again, and plan to rip each new purchase as it is made). But I also plan to encode a subset of my music as MP3s for use on my iPOD. I noticed Maxtor and Western Digital both have 250 GB hard drives out, at both 5,400 rpm and 7,200 rpm I believe. These seem to offer the lowest cost/MB. From my preliminary reading, it seems that higher rpm is good for access times, but bad for noise. Does anyone have any view on the trade-off here? Also, I've read that anything higher than 60GB requires another plate and therefore increases noise. How much should I be concerned about noise if I'm running my WAVs through an external DAC and then through my Bryston (off-topic: any advice on external DACs with either Firewire or USB connectivity for this purpose?).

Finally, any things I should watch out for before I start this? I've used EAC and RazorLAME in the past, and have occasionally had trouble with tagging and batch encoding (but that was around a year ago). Ideally, I'd like to systematically rip my CDs, then batch encode the ones I want to transfer to my iPOD. Tips on the best way to do this (or pitfalls to avoid) would be much appreciated.

Btw- sorry if this should be the start of a new thread or if some of the q's belong in other categories. I'm a newbie (first post) and am still trying to figure out proper etiquette, protocol, etc. Please trust that I HAVE spent a considerable amount of time searching around this forum before posting these questions. Anyway, feel free to suggest other places I should post this or threads I should check out to get answers. Thx.

If you're storing your CDs losslessly on disc for quality purposes, you might as well use lossless compression like FLAC and save space. Encoding from a FLAC will result in identical files as encoding from the uncompressed Wav, and you can always turn the FLAC encode into the original Wave with a minimum of fuss.

If you have an iPod, AAC would be a good choice

All else being equal, a 5400rpm drive will be quieter than a 7200 rpm drive, but often the noise difference isn't all that substantial, and the price difference is tiny. Most 250GB models these days use three 80GB+ platters, and come in around 240GB when formatted. The main advantage of the 7200 rpm drives is that the warranty period of the drive is longer- the $255 250GB 5400 rpm Maxtor IDE drive has a 1 year warranty while the $265 250GB 7200 rpm version has a three year warranty.

@rsp22 if you want a quiet hard drive (or a quiet anything computer-related), Silent PC Review is the place to do your research. Some of the quieting technology available for 7200RPM drives is making them quieter than their older 5400RPM predecesors. Some models to look for (quietest first, the rest roughly in order): Seagate Barracuda IV, Samsung Spinpoint 1204N/1614N (or some derivative of those model numbers), Barracuda V, Barracuda 7200.7, Hitachi 180GXP, Maxtor DM+9. That's from my second-hand experience, and it's also what the site admin recommends from his in-depth experience. The WD drives are screamers, I've heard, and so are the ball-bearing Maxtor drives, which I can claim from personal experience. I heard a fluid-dynamic-bearing Maxtor drive recently, and it was much, much quieter.

@Audible!: it sounds like you know more about this RAID stuff than I do, so I'll accept your judgement that RAID 1 would be more appropriate than RAID 5 for this situation. My reasoning was that, with 3.5TB (about 20 drives), it's cheaper to tack on a few extra drives in RAID 5 (it is possible to have, say, three redundant drives in a 23-drive RAID 5 array, correct?) than to buy 40 drives and use RAID 1.

@Mac: Okay, I'll re-do my calculations. Let's say 12 minutes per CD, because he won't be attentive at the ripping station, and 5/60ths of $30 an hour. It will take a year of 40-hour work-weeks to finish the job, and still be worth $5000 of one's time.

Edit: Audible!, I think your comment is a result of my poor use of terminology; I should have said "multiple parity drives" or something along those lines, rather than "redundant" which suggests duplicate drives ("RAID 5+1"?). Does my suggestion now sound reasonable?

I thought of a good solution. Hire a team of high-school or college computer geeks to do the job for you. Tell them that they can keep copies of whatever they want, and they'll do it for free. Pay them a nominal sum to keep copies of everything, and you'll have a backup solution. Let's say they copy a CD every 15 minutes for four hours after school on Monday, Wednesday, and Friday. If you hire 20, you'll have the job done in 2 1/2 months.

I believe the ratio in RAID 5 is usually 1:2, meaning one allowable failure per three total drives (and ~2/3 capacity of total drives), but I'm unsure about very large arrays. Edit: Actually, I'm pretty sure this is wrong - the minimum amount of drives for a RAID 5 array is three, and this can withstand a single drive failure, but I'm unsure how this scales.
This doesnt make RAID 5 more resistant to surges or failure though, and an array will take longer to rebuild in the event of disk failure.
My concern was primarily with price - 20+ large hard drives just isn't cheap! This is why lossless seems untenable to me for someone with a very large collection. MPC can sound really excellent, and it's small enough to be usable here.
I'd use MPC primarily if my portable and car players supported it. What I was suggesting was a small mirroring array for lossy compression because a lossless array would be both huge and expensive in comparison.
He could use a RAID 1 or 5 array with 5-6 250GB drives and it shouldnt cost more than ~$1900, controller included, and he wouldnt have to fear much assuming the host computer is on a surge protector. He could even pull the mirrored drives and rebuild, then store a set in a safety deposit box if he felt like it.

For a lossless array, you're talking closer to $6500 on drives alone, and good luck finding a controller that supports more than 16 drives for even that much by itself!
$1,900 vs. $10,000+ is a big difference, especially for a matter like this.

Here's a cheap solution to the 10K CD dilemna:

Sort through them and sell what you don't listen to. Or, just rip what you do listen to and when the mood strikes, the rip the next in line. 10,000 CD's is a number that's just not possible to listen to in any reasonable length of time. I'm inclined to say you'd be wasting time and money in trying to rip/encode all of them.

I'll go hide now.

Prices have come down a bit. Here's a real world pricelist of what you'd need:

3WARE Escalade 7500-12 $519.00 x 2 = $1038.00
Hitachi Deskstar 180GXP 180GB $155 x 24 = $3720

That's $4758 for two RAID5 arrays of 12 drives each. Total capacity is 3960GB (3600GB if we use 2 drives as hot spares). Add $1500 for a proper case/ps/mainboard/etc.

Grand total: $6258

Not bad for roughly 4TB of storage.

It will take years to rip listen to those 10,000 discs. Why Not upgrade harddisks over the next couple of years and rip some 400 at at time

I think he's more interested in archival than just listening. I.e., to avoid losing part of his music investment if CDs get lost, scratched, etc. 10,000 CDs is a LOT to keep up with and take care of...I can't find CDs in my collection half the time, and I have less than 400!

But I agree with your point on pacing the project over time mainly because of storage technology prices dropping in the months and years to come (per my post about ten or twelve back in this thread).

I'd have to have a pretty compelling reason to spend thousands on hard drives, controllers and supporting components all at once, and then in 18 months (or less)find that equivalent storage capacity costs a small fraction of what I paid.

I'm looking forward to 50 GB Memory Sticks, myself...

------------------------------------------------------------------------------------

2005: 10 GB Memory Sticks. 50 GB flashcards. Hard drives at 500 GB per platter (after a kid in a garage discovers a hot new magnetic-media indexing algorithm).

2007: Hard drives at 2 TB per platter (the kid's a billionaire by now). 500 GB flashcards.

2010: 10 TB on a standard home PC. 2 TB flashcards. Garage-kid from five years ago invents a 100 GB experimental, upgradable cranial implant with integrated neural interface (but anyone who has one will be considered a freak). Digital media compression still needed to store 20TB-worth of high-definition media on your 10TB PC.

2015: 500 TB on a standard home PC. No more spinning disks: 100 TB per flashcard, with 5 to 10 flashcard slots on an affordable computer. ALL PCs are portable now. 10 TB now available for the no-longer-experimental cranial implants (anyone who DOESN'T have one will be considered a freak). Cranial media storage coupled with thought-controlled satellite-commlink implant device...cell phones are obsolete. Digital media compression is a thing of the past...now the catch phrase is *throughput optimization*. Brain surgeons become digital storage media gurus.

2020: 100 Exabytes (100,000,000 TB) on a chip hanging from your keychain (with a built-in headphone port, of course, for those times you don't want to use the 500 TB chip in your head to listen to music from). "Storage media wars" have given way to "bandwidth wars" as the atmosphere is permeated with radio and microwave communications over every available frequency range, with people fighting over a little more bandwidth for themselves...

...and the march goes on...

Is the 10000 CDs guy still reading this thread even?

Sidebar: Why am I getting the impression that people like playing with their computers more than listening to their music?

Dex

Actually, right now I'm:

1. Ripping a CD.
2. Encoding four files at a time with LAME.
3. Running MP3Gain on another album.
4. Writing a document.
5. Using the internet (obviously).
6. "Playing with" some other things on my computer.
7. AND listening to Milla Jovovich...

...all at the same time. Ya gotta love multi-tasking...

lol. I'm your age, and my collection size is SQUARE-ROOT yours.

re cost, If the guy has (had) between $100,000 and $200,000 to spend on the CD's in the first place, then several thousand for hard drives shouldn't break his bank.

OTOH, the people I know with large collections of original (vs pirate copies) music get a lot
at garage sales, estate sales, etc. One guy I rented a room from had at the time around
7,000 hours worth of records (before CD's came out). Imagine encoding all that with the
added labor of splitting out and tagging the tracks.

Given the enormity of the task, the original poster (to the extent he/she is still following this) may want to consider outsourcing the job. The AudioRequest people offer a Music Loading service, though it's unclear how much it would cost or if it's available if you don't purchase one of their (arguably overpriced) systems. Also, they sell RipStations for people with similar needs. Not sure if this is what SometimesWarrior was referring to when he talks about "ripping tower and encoding cluster". Neither solution is cheap, but more than one post has already demonstrated that cheap (in absolute terms) isn't an option and instead the key is to find a solution that is RELATIVELY cost/time-efficient given the requirements. Still not sure if AudioRequest qualifies, but if you wanted to learn more: http://www.request.com/us/

Regarding the future, I don't know if I agree with the ScorLibran's timeline for cranial implants, but the holographic storage that he/she speaks of IS just around the corner: http://www.inphase-technologies.com.

Audible!: Lossless would seem like the best answer for me (FLAC seems to be getting the most support- see [overpriced] AudioRequest device). A quesiton on lossless: Is there any decoding time required in playing lossless files that is not an issue for WAV files? If not, is there ANY advantage to WAV whatsoever?
Regarding AAC, I've got the "old" iPod for Windows- I bought a 20GB windows-compatible iPod last september straight from the first run of the assembly-line only to see it made obsolete with the 2003 iPod!! It's my understanding that mine doesn't support AAC. Even if a software upgrade could fix this, I'm worried about ubiquity of AAC vs. MP3 and also just don't feel like going through the hassle of educating myself on optimal AAC bitrates, etc. (I've spent a not-insignificant amount of time researching the optimal lame settings for MP3). That said, I'd be interested in people's thoughts on this decision point.

Well,if you got a 10000++ cds..Even if you bought a blanks and burned-copyed from friends??its a considerable cost......So what is a hd space compared to??On the other side,back to your question i would go with a mpc(q5) for a majority of them,and for cds you really like go with q6-7 or losless for those--in case you need to convert them in mp3s..like for portable mp3 player and such.....

Of course there's more decoding time for encoded music vs wav. That's kind of the point

e.g. Annie Lennox - Bare - The Hurting Time, timed using foobar2k's speed meter (foo_null.dll):

MP3, 224kbps: decoding took 8632 milliseconds, speed 52.38x
APE, high: decoding took 18246 milliseconds, speed 24.77x
FLAC level 5: decoding took 7030 milliseconds, speed 64.30x
WAV, 16 bit undithered: decoding took 872 milliseconds, speed 518.44x

Depends if 10x the CPU's worth ~60% the space to you.

How about this http://mittelhauser.org/pdms/products.html
or this http://www.apple.com/xserve/raid/

The deal with RAID-5 is that only one drive in an array can fail and the array still survive. And the minimum number of drives is three. Yes, you could have a huge 20 drive array (you'd have to use software RAID) and you would only lose one drive to parity. If you just had 7 3-drive arrays, you'd lose 7 drives to parity. Then again, this is a direct trade-off to the fact that it's much more likely that two or more drives will fail in a 20 drive array then it wil in a 3-drive array. Most people find 4-8 to be the optimal number, but this is also probably related to limitations with hardware RAID.

I currently have half a terabyte in my fileserver and in the next month or so will be upgrading to a full terabyte. I will be using Samsung 5400RPM 160GB drives managed by Linux software RAID. I don't care how slow it is since I'm mostly concerned with just the amount of storage I can get for my budget.