Re:Other IDE devices with IDE RAID
on
IDE RAID Examined
·
· Score: 1
In most cases your IDE RAID solution will be a separate card. Even if it's on the motherboard, it will still be a separate device with its own resources. You still have two IDE channels free on the on-board controller for "legacy" IDE devices such as CD writers and DVD drives, as well as the more esoteric devices such as tape streamers and LS-120 drives.
Your RAID controller channels will invariably be distinguished from on-board (assuming an on-board RAID) by the headers being a different color. On the Soyo boards these are yellow, with the primary standard IDE being red and the secondary black.
Worth a note, but often overlooked by OEM system builders is that a channel will default to the lowest transfer rate on it's cable. For instance, if you have an UDMA133 controller and drive on the primary master and you connect a DVD ROM only capable of UDMA33 as a slave, your fast drive will be hobbled to this lower speed. Make sure, as much as possible, that you have a channel for each device, and if you have to slave, ensure you use a channel that won't affect your base performance. For two drives, use channel 0 master and channel 1 master on your RAID for your RAID set, channel 0 master for your "boot" optical drive and channel 1 master for your re-writer etc. For four drive RAID, slave the other two drives on the RAID channels.
Agreed. I'm talking business networks, not ISP-subscriber networks, though. I agree that in an internet environment or even a large university where users are users (and generally understand *what* they are using) and admins are root, *nix is ideal, configurable and flexible. However, in a business setup where each account needs access to different services, be it network shares, printers, internet connectivity, directory information or dare I say administrative delegation, Windows, IMHO, wins hands down.
Silly? I wish it were. I would happily switch now, having seen the direction in which Microsoft seems intent on dragging itself and its users. However, I am yet to see a *nix like distribution that installs across a wide range of both old and new hardware, looks the same to everyone and more importantly can be configured *without admin intervention* right from the word "go". With RIS I don't even have to join the box to the network (providing purchasing have bought the Intel or 3Com cards I specified and not tried to penny-pinch with an unbranded packet-loss generator). Joe user simply plugs his new box into the network, sticks my floppy into the drive and switches on. In about 45 minutes, if he leaves the damn thing alone and doesn't try to be "clever" he has what looks like his old machine complete with programs, desktop and email settings without me lifting a finger. I wish *nix, especially Linux, could do all this, but from my trials and experiences, the time it would take to write scripts, (excuse the french) piss about and generally wander around and configure workstations manually, not to mention re-training users, would far outweigh any advantage in initial cost.
For folk who have an interest in the sciences, pooh-poohing a scientific study, with probable extrapolation, without looking at it objectivly is probably about as un-scientific as it gets. You want empirical evidence? Here's some:
I have been trying to break from MS for ages. I can't condone a switch yet. Why? Admin costs. It isn't because myself and the other 2K admins can't understand or transfer our expertise to an archaic CLI based OS. It's not even because we can't work out how to do familiar things in *nix because we're all thick (much as you lot would have us believe).
It's actually the cost of having more staff to administer each machine. If I were to switch 100% to Linux I would have to administer each machine individually at the moment. In Windows I can simply change the group policy for whatever AD object I wish to change for, say, a virus database update or a permission domain wide for installation of a certain program. In *nix I would either have to log in to each machine myself or write a script for a machine to do it for me. Each way faffing about is a process involved. Simply put, *nix hasn't got the group and domain managment facilities that Windows has. Until it has, there's simply no competition. And yes, I love the OSS idea too, but i'm also a realist. Sorry.
do I now have to call the transparent, solid portals in my home that allow light in and me to see out? Perhaps the double-glazing companies will stop ringing me up now I can dob them in to Microshaft for patent infringment;)
Why is it most people confuse Internet with web? The www is simply one facet of the Internet even though most folks only use the www and email but even so, the dstinction still should be recognised or the Internet *will* stagnate as feared.
Mine are at any rate! A Compaq with 4X10K RPM U-W SCSI disks and an external array with five more. Sounds like a jet plane taking off as the controller starts the drives one-by-one:)
But I can see the appeal. True geeks don't give a rat's about noise. Gimme the speed!
...to put system RAM on the PCI bus. RAM needs to talk directly to the L2 cache, and in this day and age that means the processor. DDR RAM is so cheap you really shouldn't bother. If you want to do something useful with your old memory, go to a local school and upgrade one of their PCs with it!
The board you mention was probably the HyperDrive, a PCI based pseudo-hard disk drive that uses an external power supply and SDRAM as the media rather than a physical spinning magnetically coated disk. They require quite specific amounts of memory and you pay more the more they take, even if you get them without the DIMMs.
Yes, Tele2 are probably the most successful over here, but they use a licenced 3.4GHz system that can handle larger numbers of users and deliver better bandwidth. Bear in mind I was talking about a UK newsgroup here. Our implementation of 802.11b is a little different from yours, the major difference being the power restrictions and the inability to use 802.11b as a commercial venture. A lot of CLANs (community LANs) are springing up, mine included, but to avoid falling foul of the law, one must ensure that certain regulations are adhered to.
My own problem is that I can't plead ignorance. I know what 20dBm means and they know I know. That's probably the main reason I got flamed. When I suggested range could be improved using simple antennas, I immediately got the "you can't use a gain antenna" or "it won't make any difference, you'll have to drop the power" arguments. They simply didn't understand receive path optimisation over using excessive power, but the more persuasive arguments were against my use of ham licences and setting up high powered node-node links between licenced hams. The law currently does not allow for the carriage of third-party information (read "you can't get your pr0n from me" here) over the amateur network, but I argue that in a stagnant area like amateur radio, one has to try to push the boundaries and attempt to show usefulness to the community as well as the amateur "old boys" if it is to survive. Some people, however, can't stand the thought of people getting something for nothing. Even less can they stand the thought of people getting "their" spectrum for nothing and putting it to a useful purpose that shows up commercial systems as simply profiteering. You'd think they own the air. It'll be the Internet next, I swear...
There, now you've done it. You got me on politics. I better go take a break;)
This link is something very interesting for anyone who wants to make a parabolic antenna for 13cms/2.4GHz. The BiQuad design works well as a feed or as a stand-alone antenna.
In-joke alert:
Hmmm, I can see we have a non-packeteer in our midst. What makes you think that community LANs would run in this way? Some ex-packet BBS sysop would surely spit his dummy and take his 2.4GHz ball home or refuse to route to wherever because such-and-such didn't buy him a beer at the last rally...:)
Seriously, though, I got flamed up, down, left and right for suggesting such a thing in a newsgroup. Wireless networks are fine for specific applications such as point to point links, but distribution of broadband to subscribers really isn't the ideal use for it.
Speaking of Antennas, here in the UK we are limited to 20dBm EIRP (100mW from an isotropic radiator for those a little less technically minded, or a tenth of the FCC's base power) so it makes it a little harder to get that last drop of signal through. Large antennas basically help only because they create gain in the receive path. You have to pull the power down to keep within the law on the outgoing side of things, but the receive gain can mean the difference between a link and static. Makes for some interesting results when the law puts you under adverse conditions...
What you're seeing is a drive manufacturer's answer to the problem of pub commerce and rising costs. For example, "I bought 2 brandX hard drives from my mate for 50p each because he's upgraded, but I tried them both and they were duffers. I sent them back and because they were three days inside the warranty I got new ones." is a typical boast we hear from some people who are convinced they cannot have zapped the drive's logic with static, no matter how hard you try to explain the possibility. I don't usually agree with measures like these, but this time I can see where they're coming from...
There are reports (not that I advocate this, mind;) that stuffing one's drive in the 'fridge and letting it "cool its heels" so to speak, will allow you to get at your ones and zeros for a bit. Worth a last-ditch try at least...
OK, I have a Fujitsu MPG-3102AT date coded 2001-03, right smack bang where the problem occurred. It's also dead, an ex-drive, if it wasn't screwed to the drive-bay it'd be pushing up the daisies...
The problem is reported to be with the controller chip, one Cirrus Logic's CL-SH8671 batch coded 450E on mine. I contacted Fujitsu (being unfotunate enough to have purchased mine from a computer fair, silly sod) and found that they DON'T hono(u)r the warranty for end users! The b@stards! Last time I buy a Fujitsu drive.
The problem with the chip is that Cirrus, in their infinite wisdom, changed the material they use to encapsulate this huge QFP IC without telling anyone (so Fujitsu's story goes) and subsequently the reflow ovens in the SMD process were not reprofiled to take into account the new properties of the material they used. So the *chip* ended up either cooked to the point that ingress of moisture became possible during heat-up/cool down cycles or didn't reflow properly so ended up with dry joints on the legs because the new material leeched the heat away from the joints. I tried reflowing mine on an SMD rework station and no joy so I suspect the former.
Can't believe they say there isn't a problem, especially when they're rumo(u)red to be currenly in dispute with CL over this batch of ICs which they claim were sub-standard. If they were so sub-standard, how di they get through QC, humm? IS ther a QC dept.? Draw your own conclusions!
Slashdot Mirror
In most cases your IDE RAID solution will be a separate card. Even if it's on the motherboard, it will still be a separate device with its own resources. You still have two IDE channels free on the on-board controller for "legacy" IDE devices such as CD writers and DVD drives, as well as the more esoteric devices such as tape streamers and LS-120 drives.
Your RAID controller channels will invariably be distinguished from on-board (assuming an on-board RAID) by the headers being a different color. On the Soyo boards these are yellow, with the primary standard IDE being red and the secondary black.
Worth a note, but often overlooked by OEM system builders is that a channel will default to the lowest transfer rate on it's cable. For instance, if you have an UDMA133 controller and drive on the primary master and you connect a DVD ROM only capable of UDMA33 as a slave, your fast drive will be hobbled to this lower speed. Make sure, as much as possible, that you have a channel for each device, and if you have to slave, ensure you use a channel that won't affect your base performance. For two drives, use channel 0 master and channel 1 master on your RAID for your RAID set, channel 0 master for your "boot" optical drive and channel 1 master for your re-writer etc. For four drive RAID, slave the other two drives on the RAID channels.
Agreed. I'm talking business networks, not ISP-subscriber networks, though. I agree that in an internet environment or even a large university where users are users (and generally understand *what* they are using) and admins are root, *nix is ideal, configurable and flexible. However, in a business setup where each account needs access to different services, be it network shares, printers, internet connectivity, directory information or dare I say administrative delegation, Windows, IMHO, wins hands down.
Silly? I wish it were. I would happily switch now, having seen the direction in which Microsoft seems intent on dragging itself and its users. However, I am yet to see a *nix like distribution that installs across a wide range of both old and new hardware, looks the same to everyone and more importantly can be configured *without admin intervention* right from the word "go". With RIS I don't even have to join the box to the network (providing purchasing have bought the Intel or 3Com cards I specified and not tried to penny-pinch with an unbranded packet-loss generator). Joe user simply plugs his new box into the network, sticks my floppy into the drive and switches on. In about 45 minutes, if he leaves the damn thing alone and doesn't try to be "clever" he has what looks like his old machine complete with programs, desktop and email settings without me lifting a finger. I wish *nix, especially Linux, could do all this, but from my trials and experiences, the time it would take to write scripts, (excuse the french) piss about and generally wander around and configure workstations manually, not to mention re-training users, would far outweigh any advantage in initial cost.
For folk who have an interest in the sciences, pooh-poohing a scientific study, with probable extrapolation, without looking at it objectivly is probably about as un-scientific as it gets. You want empirical evidence? Here's some:
I have been trying to break from MS for ages. I can't condone a switch yet. Why? Admin costs. It isn't because myself and the other 2K admins can't understand or transfer our expertise to an archaic CLI based OS. It's not even because we can't work out how to do familiar things in *nix because we're all thick (much as you lot would have us believe).
It's actually the cost of having more staff to administer each machine. If I were to switch 100% to Linux I would have to administer each machine individually at the moment. In Windows I can simply change the group policy for whatever AD object I wish to change for, say, a virus database update or a permission domain wide for installation of a certain program. In *nix I would either have to log in to each machine myself or write a script for a machine to do it for me. Each way faffing about is a process involved. Simply put, *nix hasn't got the group and domain managment facilities that Windows has. Until it has, there's simply no competition. And yes, I love the OSS idea too, but i'm also a realist. Sorry.
do I now have to call the transparent, solid portals in my home that allow light in and me to see out? Perhaps the double-glazing companies will stop ringing me up now I can dob them in to Microshaft for patent infringment ;)
Why is it most people confuse Internet with web? The www is simply one facet of the Internet even though most folks only use the www and email but even so, the dstinction still should be recognised or the Internet *will* stagnate as feared.
Microsoft issues a statement calling the launch a "complete success" and promises 150 feet by "service pack 2"
Mine are at any rate! A Compaq with 4X10K RPM U-W SCSI disks and an external array with five more. Sounds like a jet plane taking off as the controller starts the drives one-by-one :)
But I can see the appeal. True geeks don't give a rat's about noise. Gimme the speed!...to put system RAM on the PCI bus. RAM needs to talk directly to the L2 cache, and in this day and age that means the processor. DDR RAM is so cheap you really shouldn't bother. If you want to do something useful with your old memory, go to a local school and upgrade one of their PCs with it!
The board you mention was probably the HyperDrive, a PCI based pseudo-hard disk drive that uses an external power supply and SDRAM as the media rather than a physical spinning magnetically coated disk. They require quite specific amounts of memory and you pay more the more they take, even if you get them without the DIMMs.
Yes, Tele2 are probably the most successful over here, but they use a licenced 3.4GHz system that can handle larger numbers of users and deliver better bandwidth. Bear in mind I was talking about a UK newsgroup here. Our implementation of 802.11b is a little different from yours, the major difference being the power restrictions and the inability to use 802.11b as a commercial venture. A lot of CLANs (community LANs) are springing up, mine included, but to avoid falling foul of the law, one must ensure that certain regulations are adhered to.
My own problem is that I can't plead ignorance. I know what 20dBm means and they know I know. That's probably the main reason I got flamed. When I suggested range could be improved using simple antennas, I immediately got the "you can't use a gain antenna" or "it won't make any difference, you'll have to drop the power" arguments. They simply didn't understand receive path optimisation over using excessive power, but the more persuasive arguments were against my use of ham licences and setting up high powered node-node links between licenced hams. The law currently does not allow for the carriage of third-party information (read "you can't get your pr0n from me" here) over the amateur network, but I argue that in a stagnant area like amateur radio, one has to try to push the boundaries and attempt to show usefulness to the community as well as the amateur "old boys" if it is to survive. Some people, however, can't stand the thought of people getting something for nothing. Even less can they stand the thought of people getting "their" spectrum for nothing and putting it to a useful purpose that shows up commercial systems as simply profiteering. You'd think they own the air. It'll be the Internet next, I swear...
There, now you've done it. You got me on politics. I better go take a break ;)
This link is something very interesting for anyone who wants to make a parabolic antenna for 13cms/2.4GHz. The BiQuad design works well as a feed or as a stand-alone antenna.
In-joke alert: Hmmm, I can see we have a non-packeteer in our midst. What makes you think that community LANs would run in this way? Some ex-packet BBS sysop would surely spit his dummy and take his 2.4GHz ball home or refuse to route to wherever because such-and-such didn't buy him a beer at the last rally... :)
Seriously, though, I got flamed up, down, left and right for suggesting such a thing in a newsgroup. Wireless networks are fine for specific applications such as point to point links, but distribution of broadband to subscribers really isn't the ideal use for it.
Speaking of Antennas, here in the UK we are limited to 20dBm EIRP (100mW from an isotropic radiator for those a little less technically minded, or a tenth of the FCC's base power) so it makes it a little harder to get that last drop of signal through. Large antennas basically help only because they create gain in the receive path. You have to pull the power down to keep within the law on the outgoing side of things, but the receive gain can mean the difference between a link and static. Makes for some interesting results when the law puts you under adverse conditions...What you're seeing is a drive manufacturer's answer to the problem of pub commerce and rising costs. For example, "I bought 2 brandX hard drives from my mate for 50p each because he's upgraded, but I tried them both and they were duffers. I sent them back and because they were three days inside the warranty I got new ones." is a typical boast we hear from some people who are convinced they cannot have zapped the drive's logic with static, no matter how hard you try to explain the possibility. I don't usually agree with measures like these, but this time I can see where they're coming from...
There are reports (not that I advocate this, mind ;) that stuffing one's drive in the 'fridge and letting it "cool its heels" so to speak, will allow you to get at your ones and zeros for a bit. Worth a last-ditch try at least...
OK, I have a Fujitsu MPG-3102AT date coded 2001-03, right smack bang where the problem occurred. It's also dead, an ex-drive, if it wasn't screwed to the drive-bay it'd be pushing up the daisies... The problem is reported to be with the controller chip, one Cirrus Logic's CL-SH8671 batch coded 450E on mine. I contacted Fujitsu (being unfotunate enough to have purchased mine from a computer fair, silly sod) and found that they DON'T hono(u)r the warranty for end users! The b@stards! Last time I buy a Fujitsu drive. The problem with the chip is that Cirrus, in their infinite wisdom, changed the material they use to encapsulate this huge QFP IC without telling anyone (so Fujitsu's story goes) and subsequently the reflow ovens in the SMD process were not reprofiled to take into account the new properties of the material they used. So the *chip* ended up either cooked to the point that ingress of moisture became possible during heat-up/cool down cycles or didn't reflow properly so ended up with dry joints on the legs because the new material leeched the heat away from the joints. I tried reflowing mine on an SMD rework station and no joy so I suspect the former. Can't believe they say there isn't a problem, especially when they're rumo(u)red to be currenly in dispute with CL over this batch of ICs which they claim were sub-standard. If they were so sub-standard, how di they get through QC, humm? IS ther a QC dept.? Draw your own conclusions!