srk

Member
  • Content Count

    35
  • Joined

  • Last visited

Community Reputation

0 Neutral

About srk

  • Rank
    Member
  1. Frank, I'm not sure who you're directing this to... It seems there is much less traffic on SR than there used to be, and people don't seem to care anymore about this hardware. It's really, really, really frustrating.
  2. Hi all. I've gotten great advice here before. I thought I already knew the answer regarding what drive to use in my workstation, which I am now upgrading. But Fedor (see quoted stuff below) and some others suggest that perhaps my conclusions were wrong for this next upgrade. Since my original post was as a response to an old discussion, I thought I should bring it up here where it belongs. Right now, thanks to the advice of people here, I have the fastest single-user workstation around for my purposes. I see two obvious choices: add one more Barricuda 320 gig 7200.10 (I have extras in my drawer) to make 4 drives, RAID 0, or use four Raptors in the same configuration. However, I could choose an entirely new hard drive for the RAID. My Dell Precision 690 workstation (1kW power supply) should have enough power for anything. The built-in SATA controller will, of course, handle SAS drives as well, but I don't think that would give me any advantage per se. Anyway, I've been more specific below. Any help would be much appreciated. What exactly do you do with these huge data sets? Unless you keep copying it around, and actually need to access certain bits of it, the Raptor's seek time will help you. Otherwise, yes, the Raptors ARE inferior for long sustained reads and writes and have been for a bit of a while now. There are 7200rpm SATA drives now that AVERAGE faster than the Raptor is in its fastest zone. As for the little note about the SATA1 interface being a limit, you are right, that is for each drive so whoever was saying that you'll reach the limit by using RAID was indeed flat out wrong. Sorry for the delay in my response, Fedor. I do data management and analysis on big data sets (say, 1 gig to 40 gigs). This would entail things like merging data sets, sorting them, writing new ones that are subsets of the old ones. I would create new variables from existing ones. Examples of analyses I would run would be linear regression, logistic regression, and survival analysis. My main program is SAS (http://www.sas.com/). It reads and writes data sequentially, record by record. For my current raid I'm using Barricudas (7200.10). I suppose I could just buy a Raptor and time the same data task on it vs. a single Barricuda, assuming the results for a 1-to-1 matchup should be the same as for a 3-to-3 matchup.
  3. What exactly do you do with these huge data sets? Unless you keep copying it around, and actually need to access certain bits of it, the Raptor's seek time will help you. Otherwise, yes, the Raptors ARE inferior for long sustained reads and writes and have been for a bit of a while now. There are 7200rpm SATA drives now that AVERAGE faster than the Raptor is in its fastest zone. As for the little note about the SATA1 interface being a limit, you are right, that is for each drive so whoever was saying that you'll reach the limit by using RAID was indeed flat out wrong. Sorry for the delay in my response, Fedor. I do data management and analysis on big data sets (say, 1 gig to 40 gigs). This would entail things like merging data sets, sorting them, writing new ones that are subsets of the old ones. I would create new variables from existing ones. Examples of analyses I would run would be linear regression, logistic regression, and survival analysis. My main program is SAS (http://www.sas.com/). It reads and writes data sequentially, record by record. For my current raid I'm using Barricudas (7200.10). I suppose I could just buy a Raptor and time the same data task on it vs. a single Barricuda, assuming the results for a 1-to-1 matchup should be the same as for a 3-to-3 matchup.
  4. Hi all. Long time no see. I was going to replace my RAID0 Seagate Barricudas (3 of them) with 4 Raptors (Raptor already used for boot). I will lose some space, but 600 gig would still be sufficient. I thought this was a no-brainer, but I've read claims that Raptors are actually inferior for long, sustained reads/writes. This would seem ridiculous on its face. The only issue with such r/w's would be the SATA1 limit for the Raptor. But SATA1 limits will not be exceeded by any of today's drives. I have read suggestions that if one RAID0's enough Raptors, the speed limit would be reached, but that seems, again, absurd, as the limit pertains to each drive's individual connection to the bus, not to the speed of the drives RAIDed. Anyway, if some hard drive brainiac can confirm my take I would be greatful. My system is designed for extended file read/writes, as I use it for statistics on huge data sets. Thanks to the great advice I've gotten here, my system is the envy of everyone who knows anything.
  5. srk

    Backup Software

    The big problem (last time I checked) with Retrospect products is that they are extremely buggy for backing up to DVD. Other than that, it is a very fine product line. You can probably get an upgrade to the ful Retrospect backup for maybe $25. This allows much more control over what you are backing up.
  6. Well, it looks like this topic has been met with yawns all around. Still, I'll give an update. I used four balls of Blu-Tak on the pins of the plastic caddy that holds the Raptor on my Dell 690. I would say that the noise has dropped by about 80%, and one of the balls was a bit small because I ran out of Blue-Tak (I have some on order for my three other drives so I'll probably remove the Raptor and add a bit to the fourth ball). I'm actually a bit surprised at the difference it made. Of course, with three more drives, the noise will go up a bit. But a "layered defense" against noise, starting with the Blu-Tak and perhaps ending with a bit of insulation along the sides of the case, makes sense to me.
  7. NTFS. I called Maxtor once and complained about flaky behavior. I was asked if I reformatted to NTFS. I said, well, no, I am borrowing a colleague's drive, and it's FAT32. I was told this was not advised. Of course, it might be advised for Maxtor to have sent the drive already NTFS formatted, or with a statement that NTFS was recommended for NTFS-based systems. But that's Maxtor.
  8. Of course, another approach is to use sound & vibration absorbant material either inside or outside and around the box. I am currently talking with a supplier who might have something useful for this purpose. The goal is to quiet a system without reducing air flow at all.
  9. Well, Americans didn't invent the car per se (the first powered vehicle was perhaps a military tractor built in France by Nicolas Joseph Cugnot before the U.S. existed). The Encyclopaedia Britannica has an excellent synopsis of who invented what by country. http://corporate.britannica.com/press/inventions.html Thank you. It seems that the last major thing Finland invented was skiing (co-inventor) around 5,000 years ago. Let's be fair, though: Sibelius was a Finn, and he was a great composer. Sorry. I know I promised not to reply anymore on this topic. This time for sure.
  10. By the way, I have read discussion on this forum and on http://forums.silentpcreview.com/viewtopic.php?t=8240 But the problem is that such solutions seem incompatable with the plastic drive caddies that Dell uses on some of its systems.
  11. I just searched this site but couldn't find a topic dedicated to reducing drive noise and vibration. I have Googled, but the solutions I see are not really applicable to modern systems with (e.g.) 5 hard drives! For example, if you have drive or two in a large box, you can suspend them with elastic cords. But what to do with a fairly tightly packed set of drive bays? I had an idea that seemed to work, but I haven't tested it on my noisiest drive, the Raptor 150. It's really not very noisy, nor does it vibrate that much, but seek noises are annoying because the Dell Precision 690 box magnify them astoundingly. What I thought to do was use Blu-Tak around the pins in the plastic drive caddy that holds each hard drive. Currently, the pins stick into the screw holes on the side of the drive. BTW, I thought about Blu-Tak because audiophiles use it to absorb vibration and isolate components such as speakers from their stands. And it conforms to any space. I'm not sure when it starts to melt or burn, though. Still, I formatted a 7200.10 320GB drive yesterday, heard nothing outside the box, and saw no evidence of thermal damage to the putty...
  12. Absolutely, this is one way to do it. Imagine, though, if we're comparing 4 different stripe sizes and a big arrray!
  13. Oh, I'm through. Don't know who francois and qaws are.
  14. Nil, I have traveled enough around Europe and paid enough attention to America's European friends to have some firm ideas about a particular kind of sniping, yapping anti-Americanism. But boy oh boy do the Euros get mad if an American punctures European pretentions. And frankly, I am getting tired of such pretentions, and I do not expect to see them here. whiic's comments were not just in bad taste; they were absurd and stupid. whiic didn't bother to pose hypothetical time lines with and without the US; rather, he plucked a recent legal event and inferred that Americans were a bunch of easily led zombies. Thus, it is surely relevant to give instances where such "zombies" contributed disproportionately to modern life. And, if we are to compare legal events of Europe with those of the US, surely it is fitting to compare the EU Constitution with the American one (for a start). I will resist the urge to address your response line by line, but I will respond with some particulars, and I will not pretend to have a time machine. I will simply rely on what America actually did to be "mostly responsible for modernity." The US: (0) Invented the democratic republic and thus good government. We can talk about the roles of European thinkers here, but if we consider Europe, we also must discuss all of the really bad European ideas (and implementations) as well. Every modern democracy is based on the American model. (1) Prevented Nazi Germany, Fascist Italy, and Imperial Japan from returning the West to an atavistic totalitarianism. Do you think that might have been a necessary condition for modernity, or have I really offered "one of the silliest statements you have ever heard"? (2) Prevented the USSR from controlling all of Europe after WW2. Do you think that was not necessary for modernity? (3) Was responsible for all major computer operating systems that have ever existed. (And please, Linux is simply Unix.) (4) Was resposible for all major hardware platforms. (5) Oh, yes, the light bulb, the telephone, the car, flight, mass production technologies, etc, etc. The US hasn't simply been one among many players in all this. It has been, by far, the most important player. While I cannot know what an alternative universe without the US would have looked like, that is irrelevant here. (However, having just hopped into my time machine, I would predict that had the US never existed, Nazi Germany would still have arisen, but the result of WW2 would have been quite different. Try to square totalitarianism with modernity.) As for your suggestion that America gets rich at the expense of others, I would say that America has facilitated a world economic system that only a socialist or anarchist believes has not raised all boats. Incidentally, there is nothing intrinsically exploitative about capitalism. In general, I think it is unncessary to point out such facts of life to my European friends. But just as you were unable to resist addressing a posting that was weeks old when you replied to it, I am unable to resist clarifying my position. And by the way, your comment about "Intelligent Design" is a typical Euro-sneer. I will respond by suggesting that at least Americans did not invent spontaneous generation.
  15. Hi all. Yes, I've read the FAQ on stripe sizes (http://faq.storagereview.com/tiki-index.php?page=StripeSize)! It occurs to me, though, that surely there must be a way to determine an optimal stripe size for a given set of files in a rigorous way. Consider my case. I have a new system with five hard drives (thanks to the fine recommendations of the esteemed reviewers and discussants). Drive 0 (Raptor) is for booting & applications. Drives 1-3 for data (to be RAIDed), Drive 4 for backup (compressed from Drives 1-3, which I will not be filling up). I have five years of data sets, so I can estimate their size accurately. In addition, I will be sole user. My system has 2 dual-core CPUs (Dell 690). Suppose all files are precisely the same size. In this case, the optimal stripe would be 1/3 the file size. Realistically, there will be a distribution of file sizes. One immediately appealing approach would be to set the stripe size equal to 1/3 the mean file size. However, this "golden mean" does not optimize for really big files, and perhaps those are the ones that we will really care most about (do we really care if we save only a few seconds?). In addition, it does not take into account how often we actually use the files. We should be able to determine the latter empirically, though. Does the OS have an easily useable tool for this purpose? While there might be all sorts of ways of automating this, I am thinking of some kind of Perl script, as this would be cross-platform friendly. If others have some ideas, I would love to hear them. Wouldn't it be cool to have a Java stripe size calculator?