[jb33]

Not sure if its sake oil or for real, but I just spoke with an inventor on the team and he at least seems to believe in it. He's going to join us in the discussion here so read up and get ready to hit him with some tough quesations.


The protodrive takes a standard 100 GB SATA drive with a modified in drive controller and turns it into up to 2 terabytes of storage. Drives are upgradable from 400GB base with a password.


I haver no idea how this woirks- something about platter addressing is usually inefficient and they put up to 20 bits into the space usually reserved for one bit. Efficiency is tightened with each upgrade. Performance is supposed to rival high end drives with avarage seek time of 6ms.


Standard SATA interface supports RAID controller. 6 drives in RAID5 (2 Terabyte net) will run you $2500 (edit, thats $3500) - you get a free drive. One 2 Terrabyte drive costs $2000 (you get a free upgrade). (ballpark pricing)


Drives carry a five year warranty. delivery is 90 days from purchase. He said the drives are low heat to boot.


Is this for real? talk, discuss, flame. He also said anyone could take a 100GB SATA drive and upgrade it in 20GB chuncks, but their process is patented.


Here's a link to the product data sheet. remove the directory path to get to their homepage.


jb
http://www.protoscience.com/protodri...e%20retail.pdf

 

[spearce]

Its possible this is true, but wouldn't the drive builders have thought of this already? Maybe they haven't...


I think the drives do typically waste a decent amount of space on the platters, and if you combined a new layout structure (to gain some of that space back) with a compression software built directly into the drive, its possible that the drive would be able to boost its capacity by quite a bit. But there's definate downsides.


Most drives today (and SCSI drives definately) save a good percentage of the drive for bad blocks. When a section of the disk fails, the drive automatically moves the data to another section of the disk which is reserved for this purpose. This way errors can be worked around, for a while, before the drive really fails.


Due to the difficulty to build high quality platters that are 100% error free, most drives ship from the factory with bad blocks already present, and have already "moved" those blocks to the reserved section. So you don't even have the full reserve section available on a brand new drive.


These guys might be making use of a larger segment of the reserve area for normal data, in an effort to get space back. They might also be cutting down on the checksum space the drive uses to determine if a block is bad or not. This could be bad, as the disk could be more susceptible to failure.


Interesting if its real though. Not sure I agree with the software password upgrade strategy, why not just sell the drive outright for the full price as a 2-TB drive? I guess they want a higher profit margin on the larger disks, and users don't alll want to lay out a big amount of cash for a drive bigger than what they need today.

 

[Mr.Poindexter]

If they use lossless compression on the drive, it will affect performance unless they can do it in hardware in real time. Even then, it won't help with DVD and APE music files, as they are already compressed and won't compress any further.

 

[jb33]

spearce,

hopefully Kyle will be here soon to comment on bad blocks, error correction and such. These are exactly the types of questions I was hoping you guys would come up with.


I think the upgrade is great though. There profit margin actually goes down with each upgrade. You can buy the 2 terabyte drive initially and the price per GB is lower. I like it for your second reason. For now 2 Terabytes in RAID5 (array of 400GB drives) is great for me. But if I ever need to expand my capacity I can double it with I guess 6 password prompts.


Apparently though there is an OS limit on volume size of 2 terabytes. Can you partition a RAID array into multiple volumes?

 

[jb33]

Mr.Poindexter, I explained we'd be using these DVD and music files and he said the drives are very well suited to media. Again, I really don't understand how or why but hopefully he'll be here soon to tell us.

 

[jb33]

So the protoscience guy is gonna be up here about 1pm PST today, 1/28/2004.

 

[PSIKyle]

Hi jb33,

I enjoyed chatting with you on the phone today. Yes the ProtoDrive is real and will be widely available soon. At this point we are undertaking a pilot production run of 16,000 drives which are almost sold out. Deliveries will begin March 1, with all 16,000 being delivered within 90 days.

 

[PSIKyle]

Hi folks. I have to run to a meeting and haven't the time to figure out this bulletin board just yet. I'll return later to answer questions and concerns about the ProtoDrive. As I'll explain in upcoming posts, the ProtoDrive is not magic. Today's 100-GB drive could physically store an easy 100 Terabytes but cost is the limiting factor. The ProtoDrive's only real claim to fame is that it is fairly cost effective so that it permits capacities up to 2 TB at price levels that are consistent with today's market.

 

[jb33]

Kyle, thanks for joining us here. looking forward to your subsequent posts.


Anythign you can add to help us wrap our heads around this - or me and my head...


thanks again,

jb

 

[jb33]

Quote:

Originally posted by PSIKyle Today's 100-GB drive could physically store an easy 100 Terabytes but cost is the limiting factor.

Does anyone know anyone (a third party) who could confirm this? Hardware guys?


jb

 

[ferrisg]

Seems very odd. The only thing that would make sense would to have heads deactivated to start with and activate them in firmware with the "upgrade". Nothing physical can be done, obviously, and I doubt it rewrites the zone tables as this would not be an on the fly or intelligent thing to do in an end users system.


As a side note, zone tables are very optimized on hard disk drives. And the number of spares is generally less than 10% of the user data area on a drive. So utilizing this would not gain this type of increase, and also be rather stupid if you want any type of reliability out of the drive.


The areal density on these drives must be astronomic, or else we're talking about 6 platter drives (and even then you're looking at 300GBytes a disc, which is a rather ridiculous number). The other idea is to have a specialized ASIC on the drive to do real-time compression, but this is unreliable so you would have issues with the stated capacity.

 

[ferrisg]

Saw Kyle's reply about current drives after posting my reply. I'm sure R/W engineers or media engineers would love to hear about his solution. Also consider this drive is using GMR heads, which are now old technology. Another thought would be perpendicular recording, but this is obviously not the case here.

 

[jb33]

Quote:

Originally posted by spearce Due to the difficulty to build high quality platters that are 100% error free, most drives ship from the factory with bad blocks already present, and have already "moved" those blocks to the reserved section. So you don't even have the full reserve section available on a brand new drive.

Is this why my new 160GB drive formats to 149GB?

jb

 

[BigBreaker]

Sounds like snake-oil to me.

 

[jb33]

Quote:

Originally posted by BigBreaker Sounds like snake-oil to me.

That would tend to be my response but I couldn't help believing Kyle while I spoke to him. Could be I wanted to believe him but lets see what he has to say.


Meanwhile, fact and even theory about why it may or may not be possible could be useful to get us all thinking straight and asking the right questions.


Also, Kyle when you return, you mentioned that there are a few of the drives floating around. If possible, testimonials from folks who have seen and used them could be valuable.


jb

 

[cdissmore]

Quote:

Originally posted by jb33 Is this why my new 160GB drive formats to 149GB? jb

No, that's because the hard drive industry counts a gigabyte as (one billion) 1 000 000 000 bytes, but computer operating and hard drive formatting systems count a gigabyte as 1 073 741 824 bytes. You haven't really lost all that space to formatting, it's just a matter of how the bytes get counted, and the two industries using different accepted standards of measurement. Sort of like how CRT monitors are measured as something greater than the viewable area.


Back on topic....


Count me as a real skeptic of this whole Protodrive thing...but will follow the thread with interest. They would have to have physically developed the head/recording technology for near infinite recording density, and built it into the drive before shipping. The capacity would have to be there from day one (only to be switched on later)....which really triggers my skeptical nerve...


But, please Mr. Protodrive...lay it all out...convince me....

 

[Versa]

++Snakeoil

 

[jb33]

Quote:

Originally posted by cdissmore The capacity would have to be there from day one (only to be switched on later)....which really triggers my skeptical nerve...

I think thats the idea. Capacity is expanded when you buy an upgade key so you are unlocking capacity that was already present. I asked him if that wouldn't be pretty easy to crack and he seems to think they've got that figured out. Now as to how it works ...


jb

 

[ferrisg]

I agree with you, Carey. The only feasible way I could see this being done is to have the capacity there from the start and give access to it in firmware later. Either by locking out how many tracks can be accessed (i.e., report back you have fewer usable tracks than you really do), or by having the number of heads being locked. Either way, a firmware hack could be made to easily circumvent either of these.


The bigger question is how this areal density would be achieved. Their "datasheet" quotes something about a triple-data encoding or some such thing. This is where my skepticism kicks in.


Also, I wonder why the model number is 5400, suggesting 5400rpm to me, although 7200 rpm-like latency and access times are being quoted. Interesting choice of naming convention.

 

[Versa]

Its BS. Look at hte MTBF (mean time between failure). First of all, all major hard drive manufatorers no longer use this sepcification, secondly, they list 750,000 hours. If you do the math thats 85 years. They are saying an average drive will last for 85 years. AVERAGE. yeah right. Especially when they haven't even got a working prototype yet to even take pictures of, let alone test.


Secondly look at the avg seek time of 6ms. That is faster then all 7200rpm (around 9ms) drives and slower then all 10krpm drives (around4.7ms), in other words they most likely made that figure up. Then look at the average latency, 14ms. that is incredibly high, as a 7200rpm drive has a latency of 4.1 ms, even a 5400rpm drive has a latency of 5.55ms, and they are sayign its 14ms? that thing would have to be around a 2000rpm drive. Which of course doesn't jive with their high speed avg seek time which is a combination latency and seek.


So no, these specs were definitly made up. Its either a scam (most likely) or the marketing department made **** up to sell a nonexistant product still in design. Either way, I'd stay away from it.



Edit, yep, just did the math, a 14ms latency time would mean this supposed hard drive is rotating at 2142rpm.