• Content Count

  • Joined

  • Last visited

Everything posted by NovaTC

  1. Thank you very much for this collection. While there is a lot of bitching about Tom's, I visit it from time to time and find it not that bad, too. I will look into these reviews. However, since I don't need the drives right now (capacity-wise, I could use a safer placer for my data, as traditional backup media are either too expensive or just uncomfortable), I might consider waiting for the next generation of Seagate and WD (which have mentioned in some postings in this board). The T7K250 is actually considered "next" generation, with 125GB/platter, isn't it? And then there is the upcoming "perpendicular" technology... the best time to buy hardware is always next month. BTW: Regarding perpendicular writing, if you are in for a Flash-flash, follow this link: Big weired flash animation by Hitachi GST
  2. I consider finally putting that idle SATA RAID controller on my trusty A7N8X Deluxe Mobo to some good use, by buying two drives in the 250GB range and configuring them in a RAID 1 (Mirroring) configuration. The controller is a Silicon Image Sil 3112A. Does anyone have any (good or bad) experience with this particular setup? Are the drivers usable? Is Linux supported? Is RAID 1 noticably slower than a single disk? Is the total setup actually reliable? (Since I want to protect my data from faulty drives, not put it at jeopardy). And what drives should I buy? I currently consider - DiamondMax 10 250GB (16MB cache) Unlike some writers in this board I haven't had any issues with three different Maxtor drives so far (27GB, 80GB, 200GB). -Hitachi T7K250 (8MB cache) Unfortunately no reviews so far, but since it seems to have higher data density, I asume its STR is better than maxtor. - Seagate 7200.8 250GB My last Seagate drive was a 450MB (read: MEGAbyte), which served me well till I sold it off. But I am lacking more current experience. While I have a 120GB WD1200JB (PATA) in the same system which will receive the RAID1, I do not consider the WD2500JB... I can't actually give any specific reason, but IIRC the drive series is quite dated. And I don't like Samsung for some strange gut feeling, but maybe someone can convince me...
  3. @qasdfdsaq: Yes, there _are_ reviews, I should have written "reliable reviews". :-) Which reviews would you recommend? Storagereview doesn't semm to deliver, they have some catching up to do once they get that new testbed online... . And then there is the problem with reviews per se... a review in a very reliable paper magazine got completely different rankings for drive STR, noise, access time etc. than Storagereview... so whom should I believe? The 7200.8 is not a lot different from the competition, insome benchmarks it is a tad slower, in others a bit faster. So in the end, performance is hardly worth considering, its more the soft skills like low noise... but even there, idle different benchmarkers get completely different rankings. where e.g. the Hitachi's idle noise is considerably lower than that of Seagate. However, they don't measure in decibels but use "Sone", which is thought to better represent human perception of noise. Where Storagereview says Seagate is about one dB/a quiter than Hitachi, its 1 sone vs. 0.6 sone in the mag. The 5-year warranty is actually a very good reason for buying Seagate... they obviously trust their drives more than the competition does. @alexkraemer: I don't plan to boot from the RAID1, it is for data only. Is this still a problem then? Is it possible to dual boot Linux and Windows without corrupting the array?
  4. Hi! I am looking into building a cheap RAID 5 for my home rig. If you are interested in my motivation, read on, otherwise skip to the asterix. :-) It all started when I considered buying a new hard disk to get some capacity for all those... digicam shots, you know. ;-) I had quite some trouble choosing the most reliable drive. Of course I checked the reliability survey here on SR, but the number of reports is still quite small for some of the drives I am interested in. So I thought "Why not put that SATA-RAID controller on my Mobo to good use and do some mirroring?" This would ease the fear for a loss of my valuable data (is it only me, or is it quite impracticable to backup 200GB hard drives using 4,7GB DVD-Rs?) and make the choice for the "best" hard disk a bit easier, since reliability would not be that much of an issue (unless I get unlucky and both drives fail almost at the same time). Thinking again, I didn't like the idea of loosing 50% of all capacity in a RAID1, so I checked prices for RAID-5-controllers. Prices vary widely, I could not find any good reviews so here I am, asking the experts out there. :-) To save me some trouble when routing the wires in my tower case, and of course as it is the latest technology, I would prefer SATA over PATA, but it is not an absolute must. The only hard disk I owned, that ever died on me is a 40GB Toshiba notebook drive. It bit the dust after running in 45°-tilted position for almost 1.5 years of an average 14 hours power-on hours a day. Since unlike the Toshiba I normally treat my desktop drives well (cooling-wise) I don't want to take any chances. BTW: I am well aware that RAID5 won't protect me from file system/OS errors, viruses, or my own stupidity when handling files. * The prices of RAID-5-capable (P/S)ATA-controllers vary widely, some are below 100€ (about 120$), like the HighPoint RocketRAID 1640, but as it seems it doesn't do parity calculations in hardware, leaving the job to the host CPU. Not good. Adaptec Serial ATA RAID 2410SA is about 250-300€, as is LSI Logic MegaRAID SATA 150-4 controller. Both of these bring 64Megs of cache and appear to be native SATA, i.e. no nasty Marvell PATA-2-SATA converters (at least not apparently visible on the photos of the boards I found). However, for my purposes, 300 bucks is a tad too much, in addition it is hard to get a hold on the LSI controller. So I searched on and stumbled across the Promise FastTrak S150 SX4. Its only about 160€, has a hardware "XOR-Engine", comes without cache, but with a slot to put a PC133-SDRAM into. I got one of those lying around anyways, so I could save about 100€ here. Alas, as it seems, Promise only took its old ATA-RAID-5-board (SX4000) and added Marvell converters to jump on the SATA bandwagon. I am not sure how much of an issue this is in practise (i.e lower bandwidth, higher latency, compatibility issues with certain drives?), but it will definitely loose me Native Command Queueing, which alone wouldn't be that bad, as neither LSI nor Adaptec nor most of the current drives seem to support NCQ anyways. 3Ware is recommended in some postings in this board. I have not heard much of this company before, but its 3ware Escalade 8506-4LP adapter fits my needs, but is almost 300€ aswell. From the images alone I cannot tell whether it uses PATA-2-SATA-conversion, but the specs do not mention any cache (neither cache already included nor added using a free SDRAM socket like the Promise controller). As I understand, a cache is almost a must, and therefore I would have to go for the 3Ware Escalade 9500S-4LP, which includes 128MB, but approaches ~350€. A bit pricey. So which of the adapters I mentioned (3Ware, Adaptec, Promise, LSI) would you choose when it comes to price/quality ratio? I do not need the ultimate performance, any benefit compared to a single drive would be a nice add-on, but my main concern are drive failures. Troublefree operation under common operating systems (Windows, Linux) is very important, I don't wanna loose any data to crappy drivers. As I still have unused RAM lying around, I would prefer the Promise controller, as it is considerably cheaper. Just how bad is its PATA-2-SATA conversion? What about its reliability, driver support etc.? Should I consider PATA instead? The drives and controllers are minimally cheaper, but are not really future-proof. Some of the controllers are 64Bit-PCI. My board only offers plain-old standard 32Bit-PCI, but if I understand correctly, a 64Bit-card will still work in a 32Bit-slot, unless the extra pins do not colide with any protuding elements on the board. Is this correct? Buying the "right" drives, is another issue, here I am equally undecided. As I am opting for a minimal solution for starters, I plan to go with three drives, which should also save me some noise and heat issues (my case is well ventilated though) - When it comes to performance, the Hitachi 7K250 appears to be the current king of the hill. Knowing of IBMs bad reputation with its Deathstar-series, I am a bit uneasy going with former-IBM-now-Hitachi-drives, but RAID5 would let me sleep easily anways. Unfortunately, the 160GB variant (at about 85-90€) is in short supply by all my favourite web shops, while the 250GB variant is available but considerably more expensive when comparing €/GB, costing more than 150€. - Seagate's Barracuda 7200.7 SATA 200GB (ST3200822AS) seems interesting, good €/GB-ratio (its about 110€ here). Seagates have a record for being cool and quiet (which is confirmed by the "hard drives 2.0" review here on SR), while not performing as good as Hitachi. - What about the Maxtor DiamondMax 10 160GB drive (6B160M0)? I have two Maxtors (one 27GB, the other 80GB) still going strong in my older desktop, what about performance, heat, noise? -Western Digital? The current 250GB version is quite a screamer (in the true meaning of the word) according to the SR review. Its not that fast either, still an option (in its 160GB or 200GB variant)? I am no big fan of Samsung, I met some Samsung hard disks that had gone belly up back in the hard disk dark age (2GB). Current models appear to have quite a record in terms of reliability and noise, should I consider them instead of the above drives.
  5. I plan buying a DiamondMax10 (200GB) with Serial ATA interface. Do these drives use the "old" standard power connector (like that found on optical drives and PATA-HDDs) or do they use the serial ATA power connector? If the use the SATA type, will an adapter do the trick, or do the drives use the extra 3,3V line?
  6. Thanks for the info, I will order an adapter, since my power supply does not have SATA power connectors. I just hope the drive does not use the 3,3V supply on the connector, since AFAIK these adapters only supply the standard 5V and 12V lines.
  7. Due to some legal issues that arise with the ongoing lobbying of certain hell-bound industry branches, I consider some serious large scale file/disk encryption solution. I am not talking about a few hundred megabytes of word files and family pictures, but more in the hundreds of gigabytes ranges. So if someone ever came looking on my computing for something that might put my head at risk, they would be out of luck without the proper encryption key. Since torture is still not legal in most parts of the so-called civilized world when it comes to extracting encryption keys/phrases from suspects, such a solution could save quite a few butts and bucks out there. Target platform is primarily Windows, but I might consider building a Linux file server some day. My most important requirements are: 1) reliability 2) security 3) moderate performance hit in day-to-day operation 4) seamless integration into the OS, ideally the solution should ask me for the volume password once after startup and make the data available until the system is shut down, or perhaps after a certain (adjustable) time of user inactivity 5) (hard to satisfy): accessing the encrypted volume from Windows and Linux, of course using a both worlds file system, i.e. FAT32) I checked truecrypt (, which is an open source Win2k/XP software. It offers several encryption algorithms (Blowfish, Triple-DES, AES, CAST), can work on container files as well as use an entire (empty, unformated) partition or device as container. I did a coarse test of performance on my notebook. Copying a 700MB file from the notebook's HDD (HGST 7K60) to another directory on the same drive takes about 75 seconds, with CPU usage in the <5% area all the time (yeah, not that fast, but remember, same disk, quite full and somewhat fragmented). Copying it to a 1GB container file residing on the same disk took exactly the same time. CPU usage was however about 30%, so playing media files from an encrypted drive could be a problem. I plan to do more thorough testing, but the first results regarding performance are promising. Having both the container file and the source file residing on the same disk might however be an ideal situation, since drive speed is definitely the limit. The target machine will have a faster CPU, but hopefully MUCH faster drives (RAID5?) as well, so CPU usage might become an issue. If anyone has any real world measurements, I will be happy to hear them. Testing security is not my expertise however, I will have to rely on experts regarding the principal security of the supported algorithms. The writers of Truecrypt claim that their implementation is safe, e.g. they do not store passphrases in swapable memory. The containers do not even contain magic bytes that might give away that they are actual encrypted data, they say they appear as completely random data. (termed "plausible deniability") All kinds of file systems available to windows are supported, the volumes are mounted as drives that appear "physical" to common tools, i.e. they can be formated, defragmented, recovery tools should work as well. Donwsides: no boot support (and finding a software named "TrueCrypt" on a windows install will ruin the "plausible deniablity" somewhat...), but that is not much of an issue. for my purposes, as it is data not apps or the OS I am concerned about. Seems perfect so far, still it can't satisfy 5), nor can I give any estimates regarding 1). I believe an encrypted volume is much more succeptible to damage from a few bad (physical) sectors, perhaps killing ALL files in the volume ("avalanche effect", anyone?) I plan to put it to the test by using a truecrypt "drive" as temp space for a certain software. I would like to hear some suggestions for drive encryption software, especially experience regarding Truecrypt. Greetings Nova
  8. @mohawk: What you report worries me a little. I did not expect stellar support from a budget label, but still... . I hope the nForce3 problem you report does not translate to some nForce2 boards as well. You mention a conflict with the onboard serial ATA controller on the Supermicro board. How did you solve it? Did you disable the onboard controller or did you find a workaround? About RAID expansion... you might be able to increase the size of the existing partition with a tool like partition magic (which, at least in the last version I remember did most of the dirty work under plain old DOS, so you are probably out of luck if Promise does not supply DOS drivers for the card. But it might work under Windows as well, I always used it on a bare system without a functional OS.) Linux offers tools with a similar functionality, but I never used them. Of course, using any such tool means putting your data at jeopardy. I used PM to do some minor partition modifications (on an unRAIDed disk connected to an onboard controller), I had no problems but your mileage might vary. @VDPloeg Thanks, the reviews are interesting. As you can guess, I still have not purchased anythign RAIDish. I am no easy customer, and since my vacation is over, I might as well wait for a native card by promise (i.e. SX4 successor).
  9. I find it not too hard to imagine how data migration might work. For example, when going from a basic 2-disc RAID0 stripe set to RAID 5, e.g. by adding one disk, all that has to be done is moving one third of the data from the two existing disks to the new disk while at the same time using the now free space on the two "old" drives for parity info. From a RAID 1 to RAID 5 is comparable, just move some data to the new drive, add the parity information to the two old ones. You can do it in such a way that even a hard shut down (i.e. power switch or black out) won't compromise RAID integrity, you just have to be sure that the data is always stored somewhere, plus information about the current state of RAID level migration, so that the controller knows what it was doing last and where it has to continue migration. Removing disks is another matter. When going from 3-disk RAID5 to three -disk RAID0, you simply drop the parity info. When going two two-disk RAID-0 you have to empty one drive, by successively removing parity information on the two remaining drives and replacing data. You would hardly need any free space to do this. Perhaps some sectors reserved for RAID metadata will be sufficient (but more free space will likely migration faster) . Of course you will loose RAID5 benefits immediately after you start the migration. Whenever a migration reduces the capacity of the array, you will need to provide enough free space on the disks (and perhaps do a defragmentation and free-space consolidation on the file system level). Yet here I wonder, how the RAID controller (who should not have any idea about the filesystem on the RAID) knows which sectors of the array are empty. Any ideas?
  10. @Kakarot: Occupant mentions that Promise Linux support sucks. Did you ever try running your setup with Linux? Linux support is no issue for me right now, the system I plan to place the RAID in is currently my Windows XP gaming machine and will stay that for a while, I just need some handy and reasonably safe storage space, not 24/7 operation to serve media files to several clients at once. But I might be interested in Linux support sooner or later, as I consider building a Linux home server (for firewall/routing and file serving) if I ever find the time. @Occupant: It seems Promise supplies drivers ("kernel module" whatever) for specific Linux distros (Suse and Redhat), which indicates "binary only". I am no Linux expert, but AFAIK binary drivers are always locked to a specific kernel version, i.e. no security updates unless Promise delivers a new driver as well. Is there an open source driver available by the community?
  11. The problem with these two controllers is the steep price tag. Th 3Ware Escalade 9500S-4LP is about 330€ and for that price I expect native SATA support. LSI Logic MegaRAID SATA 150-4 would be more acceptable at around 250€ and seems to be native SATA, but is hard to come by. I know Promise isn't highly regarded here, but if it was only a minor loss in performance (i.e. throughput, not CPU load!), I would be willing to accept that. It is only about 170€, I can reuse old memory and it has hardware XOR. But if someone can direct me to (so far unsolved) critical problems with the Promise RAID5 controllers that actually compromise the integrity of the array, I *might* be willing to pay the price for 3Ware. Otherwise, while I still like the RAID5 idea very much, I will go with mirroring on my onboard controller for now and spend the money elsewhere.
  12. Forgot something, can't you edit posts in this forum? Can't seem to find a button... . "Giving it a try" is a problem. If I give Highpoint a try and decide it's crap, can I migrate the array to Promise without having to recreate it, loosing all data? I highly doubt this. This is especially a problem if the array is already filled, and if I don't have the capacity to store the files elsewhere. So I would like a solution that is an almost sure hit right from the start. No switiching controllers (what to do with the other one? Yeah, "ebay", but I loathe selling at ebay...), no crappy drivers, no drivges dropped for no apparent reason etc. BTW: Tomshardware (yeah, we all know it is crap, but better crappy numbers than none ;-) ) has a Cheap-SATA-RAID-5-controller review. Highpoint performs quite good, but good ol' Tom forgot to measure CPU load.
  13. I would really like to buy the Highpoint cause it would save me quite a few bucks. But I wonder how OS-dependet this half-soft-half-hardware solutions are. Not to forget CPU load. While Bit's calculations indicate that software RAID is faster, his observations only compare one 3Ware controller with pure software RAID. How high is you CPU load when copying large files, and what is your CPU?
  14. I am aware that the US have this doctrine, i.e. that they cannot use evidence they stumble upon by accident and that is not related to the reason for the search warrant, or evidence gathered by illiegal means. It is however not like this in every country. In my country, if they do stumple upon evidence that hints to other crimes, you can and will be charged for that as wll. Say, they come looking for warez and find your beautiful canabis green house and that cool Kalashnikow rifle in your closet, than warez will be the least of your problems. Depending on the amount of canabis,you will get quite a sentence either for drug consumption (if the amount is small, so that it might seem an apprioate amount for private consumption) or for drug dealing/production if you green house holds more than a few plants. Not to forget that gun... even if you had a gun license (which you can get if you make it plausbile why you would need a gun), in my country you still would not be allowed to have an automatic rifle, which is considered a weapon of war, i.e. not for civilian use under any circumstances. This would end you up in prison for sure. Back to the point... the "fruit of the poisonous tree"-doctrine that definitely lacks in my countries laws. Evidence that was gathered by questionable means _can_ be used against you. Not good.
  15. It is _not_ about cooperation. That would be the case if you are an (otherwise not involved) witness to a crime and refuse to help the authorities. It is about incriminating oneself. While it is not done directly (like in "yes, I did it, the skeleton is in the closet"), it is done indirectly by revealing the password. A "rogue country" will not accept that the suspect is not willing to help the authorities get to evidence against him. A _true_ democracy would have to respect it. Even if it leads to the perpetrator getting away. That is among other things what differentiates a democracy from a tyranny. It is not only about elections and free speech. Agreed. But Guantanamo is one such thing that I think, does not suit a democracy. Moving suspects to other countries where they can be tortured instead of interogated in accordance with human rights is a similar matter. (Note: AFAIK the later is only an unproven rumor) I am afraid not. While some countries come closer than my country or the US in some respect (like e.g. Switzerland), they often lack in other departments. The "degree of democracy" can oftennot be determined by reading a countries constitution. It can often only be seen in day to day life.
  16. The problem is: laws are made by humans. And humans are neither free of error, nor free of malevolence. In history, many laws have been passed that were injust, inhumane, simply "not right". Yes, it works like that in dictatorships, "communist countries" and all other forms of government that don't regard human rights very high. It just doesn't suit a democracy. It seems this discussion leads nowhere. Lets agree on disagreement.
  17. @track: Of course computer files are usually not random data. But if you want to securely erase a hard disk (perhaps before selling it on ebay), you can just write random data to it. So a hard disk full of random data could just be that. And about search warrants: judges tend to be very generous when signing them. I recently read about someone's flat being searched because he was accused of having used someone else's account to pay on a porn site. While this is certainly not something to be taken lightly, the evidence was not clear (the guy in fact _was_ innocent). Actually the attorney of the state had no clue about computers, as had the judge. They filled the search warrant. And all for about 50 euros (~60$)! So don't place a bet on a judge to protect your rights as a citizen. @pico1180: You quoted Future Shock and said that he is 100% correct. My point is: you _cannot_ conmpare a door to an encrypted disc. Cause there is no such thing as an impenetrable door. The door is there, they have a search warrant, they can ask(! not force!) you to open it, or they will open it for you, certainly leaving quite a mess. But how can they force you to reveal a password. Perhaps you forgot it. Of course, that is a lie, but what should they do? Put you in jail 'til you remember? For what, perhaps because of some hundred MP3 or warez? While RIAA and other lobby organisations would like to have jail sentences for that, the relation between the severity of the crime and the punishment should not be forgotten. Currently there is a development in many countries, caused by serios lobbying of the entertainment industry, to put severe punishment (even prison sentences) for non-commerical violation of copyrights. I say that here the balance between punishment and "crime" is no longer kept. I mean, five years of prison is less than a rapist usually gets! It can't be that you are (punishment-wise) better of when you rob your local CD store than when you download the MP3s them from the internet. And even for the most severe crimes, there are measures that simply MAY NOT be taken under any circumstances in a country that calls itself civilized and democratic. Torture, locking up people without giving them proper legal support etc. The problem ist, that if the apparent crime is severe enough, many people will happily agree to do whatever necessary to make someone speak. Should someone be tortured if national security is at stake (just think about the "24" TV-series, where Jack Bauer kills and tortures non-stop, and most viewers think that's okay)? Maybe. Should someone be tortured/locked up because he _may_ have some warez on an encrypted disk? - Definitely not. But since "we have the best politicians money can buy", we might soon see laws that allow just that. Perhaps that is the time to consider leaving a country (but go where?), not only because such a law might one day be applied to oneself, but because the fact that such a law was ever adopted shows that a country is no longer "free" or a constitutional state. Leaving a country because of questionable laws might seem extreme at first, but consider what many germans (especially those of yewish origin) did during the 1930s. They saw to what the political developments were leading to, and left the country. But unlike in the 1930s there might not be a "save haven" to go to.
  18. Unfortunately, you are right. The justice system or the government as a whole is always slow to react and adapt to new developments. And often the reaction is devastating. For most governments (even the so called democratic ones!) the internet is a very dangerous thing. People can gather information from sources that are not under governmental or corporate control. People can discuss their opinions freely. So they cut down the freedom provided by this technology piece by piece, to secure the benefices. And they cover it up by claiming it is all to "protect national security", or to "protect the youth". Pradeep's RIAA&Patriot Act example describes exactly this. We have to be wathcful, or sooner than we think we will end up in a GDR- or Gestapo-style police state, where everybody spies on everyone else. It has "happened" twice in my country (or parts of it) in the last century, but judging from the current development it could happen again, here or in any other western democracy. People just don't take their history lessons seriously.
  19. @ Future Shock While the comparison between a locked room and an encrypted drive is intriguing, a search warrant, at least to my understanding, allows them to search your house. It does not allow them to force you to do what they please, i.e. you do not have to help them or actively cooperate in any way. Especially if helping them would only incriminate yourself. Of course, you should not resist actively either. I only read some sources about house searches in my country, perhaps it is different in the US. The practise is as follows: They can ask you to open a locked door. You can say, "fine, better I open it myself, because otherwise they will break it open, costing me more in the end". Or you decide not to help them (perhaps because what you did will bring you in jail for years anyway, so a torn out door is your least concern). Then they will force the door open and find all evidence anyway. But if your locked door was impenetrable (i.e. a hard disk ecnrypted with a save encryption algorithm), they can not force you to open it. Not in a constitutional state, at least to my understanding of the word, as helping to open the (theoretically of course) impenetrable door is essentially the same as incriminating one self. Impenetrable real world doors do not exist, while there are certainly encryption algorithms that are impenetrable by today's knowledge. So you cannot compare a door and an encrypted drive. I did some coarse research regarding self-incrimination and the US constitution: appears to confirm that you do not have to incriminate yourself, but it doesn't cover the search warranty & password issue. However, although obviously concerned with New Zealand law, appears to refer to a change in law that (probably under the "anti-terrorist" hood), that confirms what you said: "Subsection 1 of the new section says that the police, when executing a search warrant, can make a specified person: 'provide information or assistance that is reasonable and necessary to allow the constable to access data held in a computer that is on premises named in the warrant.' " "subsection 4 says that subsection 3 does not stop the police from making a person provide assistance in accessing a computer, even if that computer 'contains or may contain information tending to incriminate the person'." Well, if this is true, I am truely horrified. I didn't expect that cutting down civil rights has already progressed that far. A balance has to be found between the rights of the individual and the safety of a nation. This balance has obviously long since been tipped over. *me removes new zealand from list of potential vacation targets* I will do further research on that topic, especially about how this situation is handled in the US and in my country.
  20. @ Future Shock: You MUST be kidding! Is this really possible in US law? I am no US resident, in my country, no one can be forced to incriminate himself or his family members. So they can not force me to reveal the key. If such a thing is really possible in the US, they should stop calling the US a democracy. Being forced to incriminate oneself by threat of a prison sentence is in no way better than forcing the same through torture. Could someone please tell me that this isn't true?! BTW: the Truecrypt software claims "plausbile deinability" (see my first post), that means it is impossible to tell whether a file/device contains an encrypted volume or just random data. So even if what you state is true, it should still offer some protection. I mean, what if they _did_ find Truecrypt on your system, and if they did find a partition or file full of random data. They would force you to reveal the key, put you in jail, whatever. But what if it actually only IS random data, perhaps created when the disc was securely erased by its previous owner? You simply could not reveal the key, since there is none. But they would not believe you and keep you locked up forever?! Despite some bad developments regarding civil rights in recent years (thank you again, Osama... :-/), I doubt this is really possible in the USA, as this would be "Inquisition Reloaded": "Confess, witch or we torture you to death." - "And if I confess?" -"Then we will burn you alive!" Of course they will get a search warrant if they monitored P2P traffic of "their" content, but if they only find encrypted stuff, they can not use it against the acused. They would have to rely on the logs alone (which IMO can be taken appart by a good lawyer, since anything on a computer can be forged). "In dubio, pro reo"! Or did they get rid of that basic rule of any good justice system as well?
  21. @Coco I didn't want to exagerate, it is all a private issue, so hiring a consultant is not an option. :-) <hint>Just think about guys from RIAA (or a similar mafia^H^H^H^H^H organisation, that's what I meant with "hell-bound industries") visiting your home and you don't wanna give them any evidence that might screw you big time. Paying several thousand bucks in fines and/or compensation for some _private_ copies of certain... media files... is not my understanding of justice, but the result of some serious lobbying (i.e. bribery) with government representatives, like MPs.</hint> @Jeff I read about loopback encryption in a magazine article. I will use it, if I decide to set up a Linux based file server. Since I don't have Gigabit Ethernet right now (which would not only require me to buy new network cards but also new Mobos/CPUs to avoid the PCI-bus bottleneck), for the moment I need the data directly on a Windows machine. The ~8MB/s of fast ethernet just won't cut it.
  22. I checked out that Maxtor DM10 some more. According to this spec sheet, it does not have a legacy power connector: Adapters from standard molex to SATA power are cheap, but if I am informed correctly, the SATA power connector has a 3.3V rail. So if the Maxtor expects 3.3V, an adapters won't do. Then comes the question about warranty. My favourite price-check site lists the DM10 having 3 years of warranty. However, most resellers do not mention warranty at all, I called some and no one could tell me. "Ask manufacturer" was the usual answer. Maxtor itself does not mention warranty periods of this drive on their web site, either. So I should asume the worst, i.e. 1 year only. I do not buy drives with 1 year only, as a matter of principle. I don't throw any money at a company that doesn't trust its products. Which brings me back to Hitachi. Apparently, my favourite retailer just received a shipment.
  23. Didn't read your second post when writing my answer. I have read the review that included the TX4200, and wondered how long it might take Promise to offer an RAID5-version of the TX4200 (or a NCQ version of the SX4, whatever you please ;-) ). As Murphy's Law tends to strike me hard, it will be released about a day or two after I buy the "old" one (i.e. SX4). Do I _need_ the capacity right now? -No. But I still have some days of my vacation left, so now would just fit into my schedule. Perhaps I should stop trying to maximise whatever I do/buy and go with what is available now and at a reasonable price.
  24. @JoeTheDestroyer I am aware of the issues regarding file system support with Linux. Of course I will go with FAT32 if I decide to use the array from Linux. I considered a file server, but that would be overkill, since I do not have a whole family that needs its media files served 24/7. Having a second computer running is, for my purposes, a waste of energy. I would also have to go for Gigabit Ethernet, which would essentially mean at least one new mainboard, to avoid the PCI bottleneck involved in PCI GBE adapters. @Kakarot Having used a 6,4 and a 10GB IBM hard disk, I was lucky with the Deathstars as I changed from IBM to Maxtor at that time, for no special reason. The 27GB and 80GB Maxtors are still working, although I hardly fire up the machine they reside in. Having some extra speed running OS/Games/Apps off the array (which will be the only array in my house for quite a while) would be a nice touch, but like you said, the idea of mixing data and apps is not appealing. Since I am too lazy to reinstall all the stuff on the box I will at least for the time being stick with system and apps on the WD1200JB. Regarding the drives: I read the review on the MaxLine III, which is supossedly mechanical identical to the DM 10. Its performance is quite a disappointment, and the DM 10 with only 8 MB cache should perform lower still. Maybe it was an issue with the reviewed sample, but unless we see benchmarks of another here on SR, we will not know for sure. I still wonder whether loosing the MD10's NCQ to the Promise controller is really a loss for my purposes. I saw some benchmarks (I believe it was in SR's TCQ/RAID/etc. article) that actually showed disks performing slower with NCQ in desktop use.
  25. Thank you guys, you have all been very helpful! @Bit While the thread was an interesting read, I don't wanna go with an extra machine, i.e. an external RAID. I want it all to end up in my existing box. I got enough cables running through my room already. ;-) About write speeds: I understand that I will not get the same increase in write speed as I would if I went with the Kamikaze version of RAID (RAID0, that is...), but still I should get an increase compared to a single drive. Even if this is not true, as I said, performance is secondary, protection against failing drives is my primary goal. I have just read your benchmark numbers (had to download OO first). While CPU usage is about 30% against 10% in software RAID, hardware RAID write performance appears to suck big time. Really disappointing. I wonder, what kind of caching strategie did you use? Could it be a problem with the 3Ware controller? Any links to numbers for other controllers? Still I prefer hardware RAID, since it should allow me to use the array with two operating systems, which probably will not work using software. I have no experience with software RAID5 anyways, so correct me if I am wrong. What software did you use? Reading you spreadsheet ("Bonnie", ext3,...) I asume it was exclusively Linux, I want Windows aswell (or, to be honest, only Windows for now, Linux to come eventually on that machine). I am not aware of any software RAID-5 for Windows. @Kakarot What you describe is (controller and disk-wise) very much what I plan to do. The cheaper promise controller (I only got 128Megs of Apacer PC133 lying around here, which I ripped out of my old Celery 400 that stood in the corner unused and diskless for about a year) with only three disks in the 160GB range. I am especially happy to hear that your RAID proved reliable, i.e. no software/driver problems whatsoever under the OS family I plan to use primarily. Like you, I want to consolidate quite a large number of media files in one place, instead of the curent variant of distributing them across several harddisks on three different machines, always being sure to have a copy of each file at least on two different disks. It is not only wasteful, but also very uncomfortable. I am aware of the PCI bus bottleneck. My mobo has got two 100MBit ethernet ports, one of which is built into the nForce2 chipset. I haven't checked, but I believe it is not connected via PCI. Even if I lost fast ethernet's 8MB/s, that wouldn't hurt too much. I don't need more than one media stream right now, and if I ever do, I will want a new board with CSA GBe anyways. ;-) I am almost ready to hit that order button, but I am not totally sure about the drives yet. While having a Hitachi as my primary working disk in my notebook, I still feel somewhat uneasy with Hitachi 3.5". If I just could get a hold of some serious benchmark of the latest maxtor desktop drives (Diamond Max 10). Some retailers offer them with a three year waranty, so they should not be inferior quality-wise in any aspect than the expensive MaxLine III, which is only available in 250 and 300GB variants. Since the DM 10 offers NCQ, it appears to be an all-native SATA disk, which makes it preferable over the Hitachis even if the controller does not support it . A general observation is that most people use an extra disk for the OS. As I understand, a hardware RAID 5 should be bootable, any reason for not putting the OS on the RAID as well?