If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#21
|
|||
|
|||
"Bill Todd" wrote in message ... "Meurig Freeman" wrote in message ... ... Without meaning to take sides, would someone mind explaining a situation whereby a general UPS would not be sufficient to protect data integrity in the event of a power failure? (and I don't mean things like the UPS catching fire, though I do understand this is a very real possibility). Ever seen your PC reboot on a power glitch that your more-than-adequately-sized home UPS should have ridden through without notice? Nope but then most are competent enough to select, install and configure a UPS. Next time you try it get some competent help and these things will stop happening to you. I have. And this doesn't happen only with home units: You mean ones that you've installed in your home. it has been known to happen with server-room UPSs Oh god no.... why do they let you in there? (though of course should be far less common there). The bottom line is that if you're going to use RAID-style redundancy, any non-disk elements of the system that are trusted to hold supposedly 'stable' data should be just as reliable as the redundant disks themselves. He boils these alleged rules from free space. This means that you'd need to mirror cached data in double ECC caches, using separate power supplies and UPSs (or cache-level batteries), to obtain similar levels of availability. Go take your lithium. |
#22
|
|||
|
|||
In article ,
Ron Reaugh wrote: "Thor Lancelot Simon" wrote in message ... In article , Meurig Freeman wrote: Without meaning to take sides, would someone mind explaining a situation whereby a general UPS would not be sufficient to protect data integrity in the event of a power failure? (and I don't mean things like the UPS catching fire, though I do understand this is a very real possibility). Your system power supply fails (not uncommon, actually). Some bozo knocks the system power cord loose. Someone hits the reset switch or power switch on the front of the machine, whether accidentally or through lack of understanding. The operating system crashes and reboots the machine without politely telling the RAID card to flush its cache. A UPS won't save your butt in any of these cases. Battery backup on the RAID card will. Clueless jibber. So batteries on the cached controller suddenly make the OS polite? No, Ron, a battery that's integrated with the caching RAID controller has the property that it can preserve the RAID controller's memory image across a failure of power to the system bus. An external UPS can't do that. Perhaps you live in a world in which nobody ever knocks power cords loose; nobody ever presses system reset or hard-power switches; and all other such events which might lead to loss of power to the system bus or disk drives without the RAID controller receiving ample warning are on the order of likelihood of "space aliens" and "meteorites"; all I can say is that that must be a nice world to live in, but it's not the one in which any system I've ever built lives. People are stupid; power supplies are falliable (I've lost about two or three a year, on average, for the last ten years, including two failures of the shared components of "redundant" power supplies in expensive name-brand rackmount servers); **** happens. It doesn't take a meteor strike to cut the connection between external redundant power and a RAID controller's cache. This is why, as expensive and complex and physically bulky as separate battery backup for I/O caches is, it is a standard feature of responsibly-built systems that run with write-back I/O cache. I'm sure you'll ignore all that and just spray some more drool and invective around; but don't fool yourself: all you're really acheving is to demonstrate what a clueless, foul-mouthed yutz you are for the benefit of the same readers you're desperately trying to convince. And that's a good thing; because, really, you have no right to put their data at risk. -- Thor Lancelot Simon But as he knew no bad language, he had called him all the names of common objects that he could think of, and had screamed: "You lamp! You towel! You plate!" and so on. --Sigmund Freud |
#23
|
|||
|
|||
"Thor Lancelot Simon" wrote in message ... In article , Ron Reaugh wrote: "Thor Lancelot Simon" wrote in message ... In article , Meurig Freeman wrote: Without meaning to take sides, would someone mind explaining a situation whereby a general UPS would not be sufficient to protect data integrity in the event of a power failure? (and I don't mean things like the UPS catching fire, though I do understand this is a very real possibility). Your system power supply fails (not uncommon, actually). Some bozo knocks the system power cord loose. Someone hits the reset switch or power switch on the front of the machine, whether accidentally or through lack of understanding. The operating system crashes and reboots the machine without politely telling the RAID card to flush its cache. A UPS won't save your butt in any of these cases. Battery backup on the RAID card will. Clueless jibber. So batteries on the cached controller suddenly make the OS polite? No, Ron, a battery that's integrated with the caching RAID controller has the property that it can preserve the RAID controller's memory image across a failure of power to the system bus. Oh so only part of what you said is true. We all prefer to pay attention to someone who can tell the whole truth. |
#24
|
|||
|
|||
In article ,
Ron Reaugh wrote: "Thor Lancelot Simon" wrote in message ... In article , Ron Reaugh wrote: "Thor Lancelot Simon" wrote in message ... In article , Meurig Freeman wrote: Without meaning to take sides, would someone mind explaining a situation whereby a general UPS would not be sufficient to protect data integrity in the event of a power failure? (and I don't mean things like the UPS catching fire, though I do understand this is a very real possibility). Your system power supply fails (not uncommon, actually). Some bozo knocks the system power cord loose. Someone hits the reset switch or power switch on the front of the machine, whether accidentally or through lack of understanding. The operating system crashes and reboots the machine without politely telling the RAID card to flush its cache. A UPS won't save your butt in any of these cases. Battery backup on the RAID card will. Clueless jibber. So batteries on the cached controller suddenly make the OS polite? No, Ron, a battery that's integrated with the caching RAID controller has the property that it can preserve the RAID controller's memory image across a failure of power to the system bus. Oh so only part of what you said is true. We all prefer to pay attention to someone who can tell the whole truth. No, Ron, everything I said in my earlier message was true. It's hardly my problem that you can't manage to follow the dots from A to Z. (But perhaps if you stopped playing cut-and-paste on my messages, it would be easier for you to do so!) _Because_ a battery that's integrated with an I/O cache can preserve that cache's memory image across a failure of power to the system bus, it allows the contents of that cache to survive all of the failures I gave as examples in my first message, all of which can cause data loss if you don't have an integrated backup battery _precisely because they can cut power to the system bus, and thus the RAID controller logic and memory_. An external UPS can't do that. That's why in all the common real-world failure scenarios I gave as examples -- including the OS rudely forcing a hardware reset of the machine (which can drop power to the system bus in many machine architectures), a power cord being knocked loose, an accidental or uninformed use of the system reset or power switch, a power supply failure, and so following -- and many others, battery backup on the RAID controller will save your butt, but a UPS won't. Ron, to be blunt, one of the two of us has some experience building real storage systems used in real business-critical applications in the real world, and the other of us is a notorious Usenet kook with his own alt.kook newsgroup. I'm perfectly happy to let the readers of this discussion come to their own conclusion about which of us they should believe. -- Thor Lancelot Simon But as he knew no bad language, he had called him all the names of common objects that he could think of, and had screamed: "You lamp! You towel! You plate!" and so on. --Sigmund Freud |
#25
|
|||
|
|||
"Thor Lancelot Simon" wrote in message The operating system crashes and reboots the machine without politely telling the RAID card to flush its cache. Clueless jibber. So batteries on the cached controller suddenly make the OS polite? No, Ron, a battery that's integrated with the caching RAID controller has the property that it can preserve the RAID controller's memory image across a failure of power to the system bus. Oh so only part of what you said is true. We all prefer to pay attention to someone who can tell the whole truth. |
#26
|
|||
|
|||
Jesper Monsted wrote:
Meurig Freeman wrote in : Without meaning to take sides, would someone mind explaining a situation whereby a general UPS would not be sufficient to protect data integrity in the event of a power failure? (and I don't mean things like the UPS catching fire, though I do understand this is a very real possibility). It's always safe to take sides with the people disagreeing with Ron. There are a lot of other things that can go wrong, besides the power failing. UPS, PSU or cable failures (like your cleaning lady hitting the cable with her broom) will all take out your non-battery-backed cache. In any case, i'd always go for battery-backed (write) cache or no cache at all. I am beginning to spot a bit of a pattern with the replies :-) Thanks to everyone who has replied, a lot of useful information in there - powersupply failures, accidental switch pressing, leads coming lose and all the other suggestions are indeed all real possibilites. Ultimately it comes down to price/risk trade-off, but at least now I can make an informed decision. Thaks again. -- Meurig |
#27
|
|||
|
|||
In article ,
Ron Reaugh wrote: "Thor Lancelot Simon" wrote in message [a bunch of wrongly-attributed crap] Ron, We know you're a Usenet kook and so forth, but you _could_ at least try to: 1) Include some actual new content in each message. 2) Actually attribute quoted text correctly. Alternately, you could just go talk to someone else. Have you checked out sci.physics lately? I bet you could have a great discussion with Ludwig Plutonium. Or you could go infest rec.photo.darkroom; there's a guy calling himself "Uranium Conspiracy" who seems to have a lot in common with you, too. -- Thor Lancelot Simon But as he knew no bad language, he had called him all the names of common objects that he could think of, and had screamed: "You lamp! You towel! You plate!" and so on. --Sigmund Freud |
#28
|
|||
|
|||
"Meurig Freeman" wrote in message ... Jesper Monsted wrote: Meurig Freeman wrote in : Without meaning to take sides, would someone mind explaining a situation whereby a general UPS would not be sufficient to protect data integrity in the event of a power failure? (and I don't mean things like the UPS catching fire, though I do understand this is a very real possibility). It's always safe to take sides with the people disagreeing with Ron. There are a lot of other things that can go wrong, besides the power failing. UPS, PSU or cable failures (like your cleaning lady hitting the cable with her broom) will all take out your non-battery-backed cache. In any case, i'd always go for battery-backed (write) cache or no cache at all. I am beginning to spot a bit of a pattern with the replies :-) Thanks to everyone who has replied, a lot of useful information in there - powersupply failures, accidental switch pressing, leads coming lose and all the other suggestions are indeed all real possibilites. Ultimately it comes down to price/risk trade-off, but at least now I can make an informed decision. Some intelligent folks actually do show up here. |
#29
|
|||
|
|||
On Sat, 25 Sep 2004 21:06:16 GMT, "Ron Reaugh"
wrote: "Malcolm Weir" wrote in message Ron lacks any kind of system-level knowledge or understanding, and so doesn't understand that if your filesystem believes that, say, some part of the space allocation mechanisms (File Allocation Table, for example) has been updated, but due to a controller trick it was not actually updated (merely scheduled for update, and then lost as a result of power fail), you'll end up with (e.g.) a file that occupies sectors that are also marked as free. The next file that gets created may the overwrite your data in another file, and you won't know about it until later. This wacko runs around drumin expensive RAID solutions based on wild rantings about dooms day scenarios. Errr, no, I don't. I disputed Ron's claim that the such scenarios didn't exist. Pointing out Ron's ignorance and misstatements triggers much ranting from Ron. While such situations may under some circumstances be possible Of course they are possible. And real system architects -- unlike Ron -- start with considering *all* possible failure modes. Ron operates from a "I know X, and will try to argue that weaknesses of X aren't worth worrying about". their likelihood is vanishing. That is Ron's assessment. It is not based on knowlege or experience, and as the value of the data increases, the potential loss in the event of a failure increases. Since Ron apparently doesn't deal with data that the users particularly care about, his assessment is often appropriate *for his users*. But if the data in question is irreplaceable, the equation shifts. A UPS covers the issue No, it doesn't. The fact that Ron claims it does supports the above! and a UPS is a good idea for critical systems for a number of reasons. Of course. Now, in my world, "critical" means something different, such as UPS, motor-generator sets, and alternate connections to the power grid... *and* if you cache _any_ data that the host thinks has been written, it *must* get written, which means non-volatile cache. Ignore this frequently refuted wacko. Cites to that "frequently refuted" bit? Message IDs will do! Oh, and refuted by *who*? Pointing out the flaws in Ron's argument, and then having Ron rebut based on the importance (not the existence) of the flaws, isn't "refuting" in English. Still, a simple observation should be informative: Ron: You don't need to worry about a bunch of weaknesses. Me: There exist a bunch of weaknesses; you decide whether you should worry about it or not. This is usually considered bad. Based on length observation, Ron never seems to deal with systems where it actually matters... Reality check time. Precisely. That's Ron's reality: it never actually *matters* if Ron's customers lose a bunch of data. He typically works with PC-level systems, and (at a guess) data loss is irritating and to be avoided, but if not catastrophic. Malc. |
#30
|
|||
|
|||
"Ron Reaugh" wrote in
: it has been known to happen with server-room UPSs Oh god no.... why do they let you in there? To fix stuff set up by clueless *******s like you, Ron. -- /Jesper Monsted |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
IDE RAID | Ted Dawson | Asus Motherboards | 29 | September 21st 04 03:39 AM |
P4C800-E Delux: Setting up SATA Drives with RAID | Will | Asus Motherboards | 13 | July 12th 04 04:33 AM |
How to set up RAID 0+1 on P4C800E-DLX MB -using 4 SATA HDD's & 2 ATA133 HHD? | Data Wing | Asus Motherboards | 2 | June 5th 04 03:47 PM |
Gigabyte GA-8KNXP and Promise SX4000 RAID Controller | Old Dude | Gigabyte Motherboards | 4 | November 12th 03 08:26 PM |
P4C800 deluxe mirror raid as storage only OS on SATA drive | Tim | Asus Motherboards | 0 | July 18th 03 06:37 PM |