Silver isn't cheap, but in that quantity it's not a big deal.
Silver is actually a lot cheaper than most people realize. It's about 100x more expensive than copper, or around $200/pound. But yeah, when you're dealing with the amounts used here, not even platinum is likely to have a significant impact on production costs.
We have some Areca 16 port SATA cards that we've had zero issue with. I have seen a number of postings from various people about issues and it seems to come down to this:
The cards don't work well with some motherboards/drives/backplanes/etc.
If you purchase hardware that others have working reliably, then you have a cheap/reliable/fast card. If you purchase some other random hardware, things may work fine, or you may spend days tracking down random issues and replacing with different types of hardware.
A 10 or 15K drive from a few years ago is not all that much slower than one today.
This is, surprisingly, incorrect. I haven't seen reliable statistics for the past two years, but prior to that the performance of 7.2/10/15k drives was increasing steadily. Depending on the metrics needed, over a 5 year period a drive could easily be supplanted by a drive of the next lower spindle speed and have an order of magnitude more space. In many cases this can happen in 3 years. Typically the only reason to buy a 3 year old drive model is to match drive types with currently used drives.
ZFS is certainly the most interesting solution where drive sizes have increased to the point that read errors on RAID rebuilds are practically a given. (ZFS checksums on read/write) My problem is that I have little experience in the *nix world. So setting up and maintaining a storage array where all management occurs through unknown command line options seems like a poor choice.
Are there any ZFS distributions like OpenFiler or FreeNAS where ZFS management is handled through an intuitive web interface?
Is there somewhere that lists all of the functions that are about to be deprecated? I've used ereg several times in the past few months and am going to have to go and replace those now. It would be nice to avoid the extra work in the future.
You scan the entire document in, OCR for text, and try to perform an automatic conversion. Then people go through where they see a graphic of the original scanned text and the OCR'd text. For each one the person selects it is correct, or adds the correction. I've known people that do this for a living, and this was cutting edge methodology 15 years ago. Now it is pretty standard.
The alternative method is to detect the text, but not OCR it. Force the user to enter the text themselves. In either case, you show the graphic to multiple users and check for consistent answers between them. The best part about this is that reading text can be done by minimum wage workers, and can be done extremely fast after a while.
The resulting electronic file has an image containing all lines from the original drawing, with text replacing scanned in text. This can be printed out and should look identical to the original, but with converted units.
(The conversion is not necessary to see along the OCR'd text as proper conversions are trivial to perform automatically. All you really need to see is that the original OCR occurred properly.)
I'm sorry, it really sounded like you just called the metric system a "new language fad". Seriously?
Look, it's not about saving money now, it's about the future. They've already lost a multi-million dollar project because they haven't gone metric yet, and a similar mistake in the future is not unlikely.
Much of the production for space parts occurs outside of the US, where they use what? Metric. What kind of overhead do you think they charge to supply imperial equipment as well as the metric that they supply to the rest of the world? How much time is wasted converting between yards, cubic feet, gallons, and pounds instead of using a system that mostly just involves moving the decimal point around? What about the fact that all of their scientific data is measured in metric, which is how they share it with the rest of the world? Do you really not see a reason for them to move to the new measurement system that they will undoubtedly move to at some point in the future anyway?
Besides, as the OP points out, the cost estimate for this is insane.
Yes, but I'd like to just purchase it as a single nice looking box. The manufacture of such a device seems simple enough that it should be much easier to buy than to build. Meh, I guess other people aren't looking for what I'm looking for.
I think most recent/decent motherboards have chipsets that will support multipliers fine. What I'm trying to find is a really cheap box with drive trays and a SATA port multiplier. Most of them out there either seem to have a full controller or lots of indicator lights that drive up the price.
A relatively low percentage of the population has been circumcised at any given point in history. And "lower risk" does not mean "no risk", so even if the whole world were circumcised, it wouldn't have that big of an impact on HIV.
No that's not correct. JBOD is just that. Just a bunch of disks. Has nothing to do with redundancy (or lack of redundancy).
This is incorrect. JBOD is similar to RAID 0 without striping, allowing one to use disks of dissimilar size. There are some RAID controllers that will incorrectly refer to presenting physical drives directly. However most RAID will correctly present a JBOD as a single logical volume.
Some vendors charge by the seat -- how many users do you think a payroll system for 60,000 employees has? That's right, a lot.
This is something I've never understood. In my limited experience, good software never has the client access the database directly. It will request information from a server, which queries the database, and returns the necessary data to the client. This vastly simplifies many aspects of security and reduces the "client access licenses" to just the ones used by the server.
Most university IT groups don't have stellar project managers, which is the one thing that a project of this scale (and criticality) needs.
This. Most universities pay their IT departments crap because they can get them so cheaply filled by students. Most students don't have a full academic education, and almost none have real world experience. The ones that do have the education and experience get higher paying jobs at private organizations. Having a university IT department manage a project of this magnitude is asking for trouble.
the OS should be able to do its own wear leveling.
I agree, but where does one get a RAW flash device with a SATA interface? You would think they would be cheaper as no fancy controller needs to be developed, but I can't find one anywhere.
Intel has specifically stated that the RAM cache on their drives is not used for writes. It is used while remapping sections of the drive, condensing sections, etc.
The problem with your approach is that the OS wont understand the drive as well as the manufacturer does, so it will always be a sub-optimal solution.
Do you have numbers to back this up with? I'd like to see a comparison of speed from common flash file systems and an SSD running EXT4. It can't be that hard to compare the two. It should be significantly cheaper to develop a controller to present RAW flash instead of the virtualized interface currently used.
I never cease to be amazed at how terrible the back end of Intuit's Quicken software is. There are so many freely available multi-user database solutions out there, and they haven't bothered to switch to one of them after decades of development? Truly amazing.
I work for a government agency that uses a common government financial package. The software uses a COBOL database, and requires that all users have full access to a Windows share where all of the files are kept. I'm just waiting for the day when an disgruntled employee goes in and deletes all of the files, wiping out all transactions for the day.
I think you're reading a bit far into what I wrote. I wasn't making any assumptions about your computer, in fact I didn't talk about your computer at all. I said that because XP only ships with drivers that are AT LEAST 8 years old, you will likely have quite a bit of work to do to get drivers installed for any hardware designs newer than that. However any actively maintained Linux distribution is going to have current drivers, and so is more likely to work out of box.
A more fair comparison is Windows 7, which has current drivers on the install media. And as I said, some hardware Just Works with Windows and not with Linux. For other hardware it is the opposite. It just depends.
To be fair, XP came out in 2001, that's 8 years ago. Installing a Linux distro from then would be pretty useless on most modern hardware too without a lot of internet hunting. Although Microsoft will release XP install CDs that have been updated with the latest service packs, they specifically do not add drivers to the CDs. When OEMs release a new XP CD they will usually include harddrive controller drivers to make installs easier, but that's about it.
Now, Windows 7 RC1 was just released, and has recognized all hardware I've plugged in to it from the get go. Compare this to my roommate's new Ubuntu install on 5 year old hardware and his troubles getting various things to work properly.
I do wish Linux would gain momentum in the desktop space, but it is important to remember that the desktop experience of any OS varies greatly depending on the OS as driver support is never a sure thing.
Slashdot Mirror
Silver isn't cheap, but in that quantity it's not a big deal.
Silver is actually a lot cheaper than most people realize. It's about 100x more expensive than copper, or around $200/pound. But yeah, when you're dealing with the amounts used here, not even platinum is likely to have a significant impact on production costs.
We have some Areca 16 port SATA cards that we've had zero issue with. I have seen a number of postings from various people about issues and it seems to come down to this:
The cards don't work well with some motherboards/drives/backplanes/etc.
If you purchase hardware that others have working reliably, then you have a cheap/reliable/fast card. If you purchase some other random hardware, things may work fine, or you may spend days tracking down random issues and replacing with different types of hardware.
A 10 or 15K drive from a few years ago is not all that much slower than one today.
This is, surprisingly, incorrect. I haven't seen reliable statistics for the past two years, but prior to that the performance of 7.2/10/15k drives was increasing steadily. Depending on the metrics needed, over a 5 year period a drive could easily be supplanted by a drive of the next lower spindle speed and have an order of magnitude more space. In many cases this can happen in 3 years. Typically the only reason to buy a 3 year old drive model is to match drive types with currently used drives.
Unexpected, eh?
ZFS is certainly the most interesting solution where drive sizes have increased to the point that read errors on RAID rebuilds are practically a given. (ZFS checksums on read/write) My problem is that I have little experience in the *nix world. So setting up and maintaining a storage array where all management occurs through unknown command line options seems like a poor choice.
Are there any ZFS distributions like OpenFiler or FreeNAS where ZFS management is handled through an intuitive web interface?
So, like iTunes?
Is there somewhere that lists all of the functions that are about to be deprecated? I've used ereg several times in the past few months and am going to have to go and replace those now. It would be nice to avoid the extra work in the future.
You scan the entire document in, OCR for text, and try to perform an automatic conversion. Then people go through where they see a graphic of the original scanned text and the OCR'd text. For each one the person selects it is correct, or adds the correction. I've known people that do this for a living, and this was cutting edge methodology 15 years ago. Now it is pretty standard.
The alternative method is to detect the text, but not OCR it. Force the user to enter the text themselves. In either case, you show the graphic to multiple users and check for consistent answers between them. The best part about this is that reading text can be done by minimum wage workers, and can be done extremely fast after a while.
The resulting electronic file has an image containing all lines from the original drawing, with text replacing scanned in text. This can be printed out and should look identical to the original, but with converted units.
(The conversion is not necessary to see along the OCR'd text as proper conversions are trivial to perform automatically. All you really need to see is that the original OCR occurred properly.)
I'm sorry, it really sounded like you just called the metric system a "new language fad". Seriously?
Look, it's not about saving money now, it's about the future. They've already lost a multi-million dollar project because they haven't gone metric yet, and a similar mistake in the future is not unlikely.
Much of the production for space parts occurs outside of the US, where they use what? Metric. What kind of overhead do you think they charge to supply imperial equipment as well as the metric that they supply to the rest of the world? How much time is wasted converting between yards, cubic feet, gallons, and pounds instead of using a system that mostly just involves moving the decimal point around? What about the fact that all of their scientific data is measured in metric, which is how they share it with the rest of the world? Do you really not see a reason for them to move to the new measurement system that they will undoubtedly move to at some point in the future anyway?
Besides, as the OP points out, the cost estimate for this is insane.
Most people here don't start off drunk when going to the bar. Where are you located?
Yes, but I'd like to just purchase it as a single nice looking box. The manufacture of such a device seems simple enough that it should be much easier to buy than to build. Meh, I guess other people aren't looking for what I'm looking for.
I think most recent/decent motherboards have chipsets that will support multipliers fine. What I'm trying to find is a really cheap box with drive trays and a SATA port multiplier. Most of them out there either seem to have a full controller or lots of indicator lights that drive up the price.
The CDC has a pretty good history performing studies with relatively accurate results. I think they can be considered a more source, no?
How on earth was this modded insightful?
A relatively low percentage of the population has been circumcised at any given point in history. And "lower risk" does not mean "no risk", so even if the whole world were circumcised, it wouldn't have that big of an impact on HIV.
No that's not correct. JBOD is just that. Just a bunch of disks. Has nothing to do with redundancy (or lack of redundancy).
This is incorrect. JBOD is similar to RAID 0 without striping, allowing one to use disks of dissimilar size. There are some RAID controllers that will incorrectly refer to presenting physical drives directly. However most RAID will correctly present a JBOD as a single logical volume.
Please refer to the Wikipedia article on RAID.
I think you might be looking for something like this:
http://www.amug.org/amug-web/html/amug/reviews/articles/addonics/5x1/
It takes a single eSATA connection and outputs several more. This lets you make your own enclosure or just attach naked drives.
Some vendors charge by the seat -- how many users do you think a payroll system for 60,000 employees has? That's right, a lot.
This is something I've never understood. In my limited experience, good software never has the client access the database directly. It will request information from a server, which queries the database, and returns the necessary data to the client. This vastly simplifies many aspects of security and reduces the "client access licenses" to just the ones used by the server.
Am I missing something here?
Most university IT groups don't have stellar project managers, which is the one thing that a project of this scale (and criticality) needs.
This. Most universities pay their IT departments crap because they can get them so cheaply filled by students. Most students don't have a full academic education, and almost none have real world experience. The ones that do have the education and experience get higher paying jobs at private organizations. Having a university IT department manage a project of this magnitude is asking for trouble.
the OS should be able to do its own wear leveling.
I agree, but where does one get a RAW flash device with a SATA interface? You would think they would be cheaper as no fancy controller needs to be developed, but I can't find one anywhere.
Intel has specifically stated that the RAM cache on their drives is not used for writes. It is used while remapping sections of the drive, condensing sections, etc.
Where would you buy a RAW flash SATA drive to use a file system designed for flash?
The problem with your approach is that the OS wont understand the drive as well as the manufacturer does, so it will always be a sub-optimal solution.
Do you have numbers to back this up with? I'd like to see a comparison of speed from common flash file systems and an SSD running EXT4. It can't be that hard to compare the two. It should be significantly cheaper to develop a controller to present RAW flash instead of the virtualized interface currently used.
Heh, the same for me. His post was squished up against the right.
I never cease to be amazed at how terrible the back end of Intuit's Quicken software is. There are so many freely available multi-user database solutions out there, and they haven't bothered to switch to one of them after decades of development? Truly amazing.
I work for a government agency that uses a common government financial package. The software uses a COBOL database, and requires that all users have full access to a Windows share where all of the files are kept. I'm just waiting for the day when an disgruntled employee goes in and deletes all of the files, wiping out all transactions for the day.
I think you're reading a bit far into what I wrote. I wasn't making any assumptions about your computer, in fact I didn't talk about your computer at all. I said that because XP only ships with drivers that are AT LEAST 8 years old, you will likely have quite a bit of work to do to get drivers installed for any hardware designs newer than that. However any actively maintained Linux distribution is going to have current drivers, and so is more likely to work out of box.
A more fair comparison is Windows 7, which has current drivers on the install media. And as I said, some hardware Just Works with Windows and not with Linux. For other hardware it is the opposite. It just depends.
To be fair, XP came out in 2001, that's 8 years ago. Installing a Linux distro from then would be pretty useless on most modern hardware too without a lot of internet hunting. Although Microsoft will release XP install CDs that have been updated with the latest service packs, they specifically do not add drivers to the CDs. When OEMs release a new XP CD they will usually include harddrive controller drivers to make installs easier, but that's about it.
Now, Windows 7 RC1 was just released, and has recognized all hardware I've plugged in to it from the get go. Compare this to my roommate's new Ubuntu install on 5 year old hardware and his troubles getting various things to work properly.
I do wish Linux would gain momentum in the desktop space, but it is important to remember that the desktop experience of any OS varies greatly depending on the OS as driver support is never a sure thing.