+1 Insightful. I've been using Myth since 2003 (and keeping it running using the same database from then to now is no mean feat, I can tell you - but that's the only option for people with TV recordings they don't want to delete as Myth has no way of importing random recordings). In that time the install procedure has changed from "tortuous" to merely "painfully inconvenient" - there's still alot of manual text-editing steps involved, for those of us in the UK at least, and the amount of hoops I've had to jump with for MySQL is atrocious, and anyone else who's run into the various debacles involving charset settings can attest.
I've already switched to XBMC as a frontend UI as it's superior in every way (apart from LIRC setup).
Not been a fan of netbook myself, but one thing they have given rise to is a rash of decently powered full fledged 11" laptops such as the Toshiba T110 and the Acer Timeline 1810TZ - both of which get 8-9hrs of battery life when doing light web stuff, and much lighter than the classic 14" laptop whilst still plenty big enough for "proper" work (keyboards are comfortable enough to write on for extended periods, and I have reasonably big hands). Laptops in these form factors were almost unheard of in any large numbers, especially after the death of the awesome 11" G4 Macbook.
As someone who tends to read and write on the move, I'm not the iPad's target market, but I still have to thank the netbook for getting reasonably small form factors back in the zeitgeist.
In a lovely display of quasi-reciprocal something or other, I wish I hadn't made my comment now so I could have modded you funny:) Everyone else seems to think I'm incapable of looking up the movie on IMDB and that I actually want to see it again.
Speaking of which, there should be more government conspiracy/mexican wrestling crossover movies.
If you want to read something alot more entertaining and you're happy with it being spread across multiple pages, read the pages at TV Tropes instead: http://tvtropes.org/pmwiki/pmwiki.php/Main/MagicalComputer It includes all the ten tropes in the list, plus many more, without obnoxious advertising.
It's much funnier, has exhaustive examples, and will ultimately ruin your life.
A bit more back on topic, my favourite "enhance" button was seen in some terrible movie starring Jack Black as a CIA hacker which I came across whilst, er, herbally medicated. It featured the usual "enhance" button with a (literal) twist - using "inference AI" it could turn a patchwork of images into a 3D model... including the bits that weren't filmed. The wall-banging stupidity of this was even a major plot point - the model was done so they could find out where someone had stashed the microfilm, or some such rubbish - typical modest programmers, they write their AI to infer things and it turns out to be an all-seeing eye that can observe past events witnessed by no other human. The only reason I'm sad I can't remember the name of that film is in case I accidentally start watching it again.
Run your own DNS server. Faster, and you get total control over what monkeying about is done at home.
Heck, I'm with a company called Zen internet at home, who are one of the best ISP's in the UK, and do absolutely no monkeying around with DNS (and have stupidly fast servers as well), but I still run my own DNS server.
And before anyone scream piracy, I spend at least an average of $300 on games each month, so keep me out of that endless loop.
Shut up, citizen. Our removal of options for things like dedicated servers means that in the future, not only will you be paying your $300 a month, you'll be paying another $100 a month for subscription to several online gaming services and you'll also be unable to play any of your games after we decide to end support for them 1 year after the last patch comes out, unless you subscribe to our Classic Gaming Service at only $50 a month!/still playing the original Unreal Tournament after all these years thank to dedicated servers
,blockquote>Unfortunately most people who possess these skills (valuable non-IT-related skills) don't know much about computers -- and the older, more experienced (and thus more valuable) employees tend to know even less.
Whilst I've met a fair few older people (especially accountants) who are pretty good with computers, I've one observation that rings 85% true for me - whilst older people might be much less experienced or knowledgeable about computers, they're a crapton better at admitting they don't know anything. Most of the time if something is up with their computer I'll be called on almost immediately, rather than some of the younger cock o' the walks that think because they pirated a copy of photoshop and managed to run keygen.exe they're some sort of computer guru who try and fix it themselves, usually with disastrous results (and I work in the City of London, where practically everyone has an instant +10 Level increase in Cockiness and General I'm-Better-Than-You).
I'm sure anyone here who's worked in a desktop support role will have, at one point had to explain something over the phone, or get a user to describe something to them over the phone. Given the choice between an older person and younger one who are both equally ignorant of the problem at hand, I'll take the older one every day.
Anecdata I know, but if I can't have someone in a role who knows something about computers, then I'll take a person that lets me do the job over someone that doesn't.
"McAfee Interwebs Secrutiny has detected that your outgoing mail to customerservices@mcafee.com, subject "You f**king idiotic t**tballs of a son of a ****** in the ******** with a hatstand!!!!" has been detected as Offensive Spam and will be deleted. Thank you for Trusting in McAfee! [TM]"
On a more serious note, I ran into a few small shops that were badly hit, but most of the people I know who work in the enterprise have a time delay before the updates hit the machines, which is usually a hangover from the last time $av_vendor bollocksed up an update.
Personally, I'm still a believer in most AV's being worse that the viruses themselves, and don't run any on my windows boxes - I don't think I've used a single one that hasn't fucked up at some point. Most of my colleagues feel the same way (and, IMHO, by the time it's hit your filesystem and you have that 20% chance of the AV detecting it, it's already too late anyway) and the only reason we run it at work is because of compliance issues... that and the majority of machines being a poorly patched IE6. Yay!
Ladies and gentlemen, this is your captain speaking. We have a small problem. All four engines have stopped. We are doing our damnedest to get them under control. I trust you are not in too much distress.
You're modded funny, but I have to agree. I bought my girlfriend a Toshiba T110 for her birthday, and ended up getting myself an Acer Timeline 1810TZ a few weeks back, and I get so much more use out of them than my "workhorse" HP that it's not funny.
Both are 11" screens, with CULV processors, 4GB of RAM and 250GB hard discs, and both cost in the region of £500, after I got over my gag reflex at the mention of the Acer brand - they're small, about 1.5kg each, you get 7-8hrs of web browsing out of a single charge and because of this you can usually happily go without carrying the charger around with you (an especially big deal in the UK due to the bulkiness of our foot-mangling plugs), and if you're feeling thirsty an additional battery is reasonably priced and easy to cart around. Neither of the machines were half as slow as I expected them to be either, as long as you don't try and do things like x264 encodes. I put a spare aftermarket SSD in my Acer and it's now, as they say, uber stonking and it gets carted around everywhere I do - also yay to O2 allowing me to tether my phone on a SIM-only contract.
As tempted as I was by the lowest-end 13" Macbook, it was too highly specced for my needs, and waaaay more expensive (£820 at the moment, £900 if you want 4GB RAM) whilst still delivering a lower battery life.
Not saying that apple should do anything resembling a netbook, but they have nothing in their lineup at the low end of the size spectrum (unless you count the remortage-my-negative-equity-please Air). If this announcement would have included an 11" CULV Macbook then things would have got alot more interesting, but these are still, tome, boutique powerhouses than a serious attempt at getting the foot in the door of the laptop market. Not that this is a bad thing if that's what they choose to be, but I'd like to see Apple get more mass-market appeal.
They're not faster in a number of apps; last time I looked (or, indeed, used a mechanical drive as a system disc) SSD's were the performance king by a nautical mile. I'm not aware of any type or number of hard discs that can compete on performance with even a mid-range SSD these days.
In the 90's your only option other than an array the size of a fridge was an enterprise ramdisc (and try getting either of those into your laptop). Depending on which metric is most important to you (and it should be random read/write), SSD's are pretty much an order of magnitude faster that mechanicals.
On top of that, not everyone needs the space. Corporate laptop users especially will boot off a small 30-60GB SSD and have all their documents presented over the network or over VPN. Paying $0.10 a GB is a loss when it's cheaper for you to pay $5 a GB and get $30 more productivity from an employee per day due to faster boots, shorter logins, less thrashing during the mandatory virus scan and fewer mechanical failures (which is often enough to put an employee out of action for a significant fraction of the day, or even week if they're out and about).
IIRC Intel (who make up 2 out of 5 of the SSD's I currently own) say you can write something like 80GB to one of their 160GB drives every day and not start running out of sectors for five years; I don't know of any mechanical discs (especially 2.5" SATA ones) that are capable of such punishment. Flash and SSD controllers (along with a DRAM cache and NCQ) have really come a long way in mitigating the rewrite/longevity problem.
Disclaimer: I don't understand why more people don't love SSD's, especially when the dislike seems to relate solely to their high price.
In my experience, most of the reason the "loading" bar is there all the time is because of all the slow-arsed ad and JS servers dotted all over the place, with 90% of web pages. I use Opera mostly myself, and it'll generally tell you which sites it's waiting on when loading a page - which is often handy for adding entire domains to opera's global block list.
I've ballpark tested this myself - I yanked down a copy of several sites in httrack and loaded them on the slowest web server I have access to (running on an ARM in a friends NAS), and then limited the speed of the ethernet port to 1Mb. Even when thrashing the wireless in order to add as much latency to the network as possible, the pages were still rendered virtually instantly, as all those external references also loaded almost instantaneously.
If I paid attention to driving in the movies, you need to constantly sway the steering left and right in order to stay in a straight line, and every time you declutch and shift a gear you need to do an elaborate jump-cut to a close up of your foot on the pedal, and then your hand on the gearstick. Neither of those seem like particularly safe practices to me.
In definitely-not-related-news; Novell are in serious financial trouble, and hence the possibility of losing the faith-based patent amnesty/MAD Microsoft have with Novell.
Everyone else bitching about Mono has known about this possibility for years.
If I hadn't already spammed this thread with posts I'd have modded you up - other than price/GB, most of the problems in the wiki article are exaggerated IMHO.
I don't get the issue with the difference between read and write speeds though. Hard drives are slower ar writing too, usually about 2/3 the speed of a read for sequential access - all the SSD's I've used have a similar discrepancy (thanks to clever controllers and TRIM making writes far faster).
As to the longevity issue, I guess the jury is still out on that as these things have only been in the wild for a couple of years. But IIRC Intel said you could write 1/3 of the capacity to their drives every day for five years before you start to run out of writable blocks, and even then it's meant to fail gracefully.
I don't understand the point about SATA-based SSD's being slower for writes either; even the crappiest SSD I've used (an OCZ Agility) has random 4k writes somewhere in the order of 8-10 times faster than a 10k SAS drive. Smells like baldercrap to me but I'll go give the sources a read - here's looking forward to your edit!
Most likely as a new sort of controller hanging off the PCIe bus with a generic interface for plugging in flash cards of X gigabytes. Most of the ATA protocol is geared around spinning platters, and as such has a whole load of folderol that it and the OS needn't bother with for solid state media. KISS!
And in the event your post was intended as a jest, I'd rather see it implemented as a softraid driver of USB floppies, each connected to a 1.44MB flash chip;)
Figured you were a programmer (I'm an infrastructure guy so have only a murky view of the internal workings of software) - part of the question about the apps was why two different speeds of memory would be an issue - I figured it'd be bad practice for an app to assume that it'd always be able to map memory at 20GB/s or whatever, but I don't really know the ins and outs of low(ish)-level programming.
I figured it'd work in a similar way to file cache now - a map of blocks/tables is given to the OS to use for storage, and the kernel just alternates between flagging these blocks as in or out of memory as applications demand - effectively using memory as read/write-through cache. As soon as a DMA came in, the OS would read those blocks from SSD into memory transparently, and keep them there for as long as necessary. Similarly, a file object that's opened read/write in memory and has been static for X seconds would be transparently written back to the SSD, or immediately written back to disc as soon as the file handle is closed. I figure a small battery and/or capacitors would be able to handle keeping the data cached (or even written) in the event of sudden power loss to help prevent corruption, plus I'm pretty sure you could implement journaling as well.
But again, IANAP so please feel free to pick holes in my naive assumptions:)
Agreed, and the nearline storage we implemented in my response to amorsen's post was along those lines - the "software" we wrote was about 250 lines of python and a couple of bash scripts for maintenance, with the rather awesome rsync+hardlinks doing the heavy lifting - I use a similar homebrew thing for backing up my home network, which implements (whole file) "deltas", date/time versioning and date pruning. I'd love to do the same thing at work with access to "proper" backup hardware + APIs (for things like taking consistent snapshots of DBs without dumping them to file) but the money for it isn't there.
Point is very much taken about tape reliability (although hard drives really don't travel well at all), we've had at least two that came back corrupt following a test DR. However, most of our contracts mandate tape in offsite escrow, and no beancounter wants to pay for both hard drive AND tape backup.
Indeed, yes, you're correct, alot of this is all about the terminology - I tend to conflate the terms myself, but then the distinction between the two can be very murky. True, most of the time a backup of, say, a commonly used database will be next to useless of it's more than a month or two old, but I've run into scenarios where we've had silent corruption in a (thankfully infrequently used) DB that the DBAs didn't notice for nearly six months, which resulted in us pulling the archival tape and restoring from that.
Personally, I wish we had the money for a nearline HDD backup or a VTL in my current job, but seeing as we *have* to have tape for archival reasons and we can turn most restores around fairly quickly, it's not seen as a necessary expense.
Dunno about magentic leakage as tapes aren't really my area (mainly because I hate the damned things:))
Personally, I don't really think people using a single tape drive without a robot/library count as "enterprisey" enough for my blood - if I had a single drive, I too would use a nearline HDD-based storage system to keep the last few versions close at hand so there's no need for the rigamarole of spooling through a tape to restore just one file. We implmented just such a "on the cheap" system in a previous job some years ago - we backed up to tape directly from the file server, but also maintained a backup server that rsynced the files over every hour. It used hardlinks to allow us to keep several faux-snapshots of the entire system whilst only actually keeping one instance of every unique file, giving us about 4 months of instant restore-any-file-you-like in just 2x the amount of space in use on the file server. Easy to set up and saved us a fortune on backup infrastructure.
I believe mainframes already do the whole memory page == disc block thing, but I'm not an expert.
I asked the same question on/. a while back regarding more common operating systems and got this response from m.dillon, which seems to indicate it's not really feasible unless the whole software stack is (very) radically altered, or the performance delta between memory and storage becomes alot smaller.
How much for a hard drive that's as fast as that $125 SSD?
It was probably a rhetorical question, but I'll answer anyway: lots. Lots and lots and lots.
We've in the middle of replacing 48U's worth of short-stroked fibre channel discs with 4U's worth of solid state drives. Capacity was never much of an issue with these databases (they only total about 800GB) but to get the performance with an IBM pSeries box cost stupid money - I don't know the exact figure but it was somewhere in the region of 50k a year just for maintenance.
Even if you're not using hardware like a pSeries, a SAN or disc array capable of sustaining >20,000 IOPS is still going to cost you silly money, take up alot more space and eat at least five times more power.
I reckon we'll need a couple more years of SSD acceptance before this becomes mainstream - for one thing, we'll need an OS-agnostic method of using PCIe cards as bootable block devices, which will probably take a while to work out. Heck, I'm not even sure if FusionIO is bootable yet (I think they're working on it). And then there's all those filesystems in use that all assume they're on spinning discs.
SSD will be made of awesome when this happens though - SATA has been a bottleneck for flash for quite a while now, and removing a) the slow-assed bus and b) the complex disc controller overhead and the futzing around that the OS does will give gargantuan speed improvements. I expect there'll be a shift like there was from MFM to IDE - move the control of the hardware away from the OS and into a smart hardware controller that embedded with the device. SATA is just used now because it's common and universally supported by pretty much everything - crucial at the "early adopter" stage.
Not really - whilst spinning platters are awesome for nearline storage, they're not nearly reliable enough to be used as a backup solution IMHO. Stick a bunch of caddied HDD's in a cupboard and leave them for a year and you can practically guarantee one of them will fail to spin up. Tapes will remain in the backup for the foreseeable future because, TTBOMK, there's no other media that's reliable enough over the long term.
Big-assed RAID arrays used for backup storage are all well and good, especially in a small or low-budget environment, but enterprises use tapes for a reason.
Slashdot Mirror
+1 Insightful. I've been using Myth since 2003 (and keeping it running using the same database from then to now is no mean feat, I can tell you - but that's the only option for people with TV recordings they don't want to delete as Myth has no way of importing random recordings). In that time the install procedure has changed from "tortuous" to merely "painfully inconvenient" - there's still alot of manual text-editing steps involved, for those of us in the UK at least, and the amount of hoops I've had to jump with for MySQL is atrocious, and anyone else who's run into the various debacles involving charset settings can attest.
I've already switched to XBMC as a frontend UI as it's superior in every way (apart from LIRC setup).
Not been a fan of netbook myself, but one thing they have given rise to is a rash of decently powered full fledged 11" laptops such as the Toshiba T110 and the Acer Timeline 1810TZ - both of which get 8-9hrs of battery life when doing light web stuff, and much lighter than the classic 14" laptop whilst still plenty big enough for "proper" work (keyboards are comfortable enough to write on for extended periods, and I have reasonably big hands). Laptops in these form factors were almost unheard of in any large numbers, especially after the death of the awesome 11" G4 Macbook.
As someone who tends to read and write on the move, I'm not the iPad's target market, but I still have to thank the netbook for getting reasonably small form factors back in the zeitgeist.
In a lovely display of quasi-reciprocal something or other, I wish I hadn't made my comment now so I could have modded you funny :) Everyone else seems to think I'm incapable of looking up the movie on IMDB and that I actually want to see it again.
Speaking of which, there should be more government conspiracy/mexican wrestling crossover movies.
If you want to read something alot more entertaining and you're happy with it being spread across multiple pages, read the pages at TV Tropes instead: http://tvtropes.org/pmwiki/pmwiki.php/Main/MagicalComputer It includes all the ten tropes in the list, plus many more, without obnoxious advertising.
It's much funnier, has exhaustive examples, and will ultimately ruin your life.
A bit more back on topic, my favourite "enhance" button was seen in some terrible movie starring Jack Black as a CIA hacker which I came across whilst, er, herbally medicated. It featured the usual "enhance" button with a (literal) twist - using "inference AI" it could turn a patchwork of images into a 3D model... including the bits that weren't filmed. The wall-banging stupidity of this was even a major plot point - the model was done so they could find out where someone had stashed the microfilm, or some such rubbish - typical modest programmers, they write their AI to infer things and it turns out to be an all-seeing eye that can observe past events witnessed by no other human. The only reason I'm sad I can't remember the name of that film is in case I accidentally start watching it again.
Run your own DNS server. Faster, and you get total control over what monkeying about is done at home.
Heck, I'm with a company called Zen internet at home, who are one of the best ISP's in the UK, and do absolutely no monkeying around with DNS (and have stupidly fast servers as well), but I still run my own DNS server.
Shut up, citizen. Our removal of options for things like dedicated servers means that in the future, not only will you be paying your $300 a month, you'll be paying another $100 a month for subscription to several online gaming services and you'll also be unable to play any of your games after we decide to end support for them 1 year after the last patch comes out, unless you subscribe to our Classic Gaming Service at only $50 a month! /still playing the original Unreal Tournament after all these years thank to dedicated servers
,blockquote>Unfortunately most people who possess these skills (valuable non-IT-related skills) don't know much about computers -- and the older, more experienced (and thus more valuable) employees tend to know even less.
Whilst I've met a fair few older people (especially accountants) who are pretty good with computers, I've one observation that rings 85% true for me - whilst older people might be much less experienced or knowledgeable about computers, they're a crapton better at admitting they don't know anything. Most of the time if something is up with their computer I'll be called on almost immediately, rather than some of the younger cock o' the walks that think because they pirated a copy of photoshop and managed to run keygen.exe they're some sort of computer guru who try and fix it themselves, usually with disastrous results (and I work in the City of London, where practically everyone has an instant +10 Level increase in Cockiness and General I'm-Better-Than-You).
I'm sure anyone here who's worked in a desktop support role will have, at one point had to explain something over the phone, or get a user to describe something to them over the phone. Given the choice between an older person and younger one who are both equally ignorant of the problem at hand, I'll take the older one every day.
Anecdata I know, but if I can't have someone in a role who knows something about computers, then I'll take a person that lets me do the job over someone that doesn't.
"McAfee Interwebs Secrutiny has detected that your outgoing mail to customerservices@mcafee.com, subject "You f**king idiotic t**tballs of a son of a ****** in the ******** with a hatstand!!!!" has been detected as Offensive Spam and will be deleted. Thank you for Trusting in McAfee! [TM]"
On a more serious note, I ran into a few small shops that were badly hit, but most of the people I know who work in the enterprise have a time delay before the updates hit the machines, which is usually a hangover from the last time $av_vendor bollocksed up an update.
Personally, I'm still a believer in most AV's being worse that the viruses themselves, and don't run any on my windows boxes - I don't think I've used a single one that hasn't fucked up at some point. Most of my colleagues feel the same way (and, IMHO, by the time it's hit your filesystem and you have that 20% chance of the AV detecting it, it's already too late anyway) and the only reason we run it at work is because of compliance issues... that and the majority of machines being a poorly patched IE6. Yay!
I'm pretty sure the eruption was caused by Steve Jobs trying to hollow out the volcano. Iceland is the perfect place for any supervillain lair.
http://en.wikipedia.org/wiki/British_Airways_Flight_9
You're modded funny, but I have to agree. I bought my girlfriend a Toshiba T110 for her birthday, and ended up getting myself an Acer Timeline 1810TZ a few weeks back, and I get so much more use out of them than my "workhorse" HP that it's not funny.
Both are 11" screens, with CULV processors, 4GB of RAM and 250GB hard discs, and both cost in the region of £500, after I got over my gag reflex at the mention of the Acer brand - they're small, about 1.5kg each, you get 7-8hrs of web browsing out of a single charge and because of this you can usually happily go without carrying the charger around with you (an especially big deal in the UK due to the bulkiness of our foot-mangling plugs), and if you're feeling thirsty an additional battery is reasonably priced and easy to cart around. Neither of the machines were half as slow as I expected them to be either, as long as you don't try and do things like x264 encodes. I put a spare aftermarket SSD in my Acer and it's now, as they say, uber stonking and it gets carted around everywhere I do - also yay to O2 allowing me to tether my phone on a SIM-only contract.
As tempted as I was by the lowest-end 13" Macbook, it was too highly specced for my needs, and waaaay more expensive (£820 at the moment, £900 if you want 4GB RAM) whilst still delivering a lower battery life.
Not saying that apple should do anything resembling a netbook, but they have nothing in their lineup at the low end of the size spectrum (unless you count the remortage-my-negative-equity-please Air). If this announcement would have included an 11" CULV Macbook then things would have got alot more interesting, but these are still, tome, boutique powerhouses than a serious attempt at getting the foot in the door of the laptop market. Not that this is a bad thing if that's what they choose to be, but I'd like to see Apple get more mass-market appeal.
They're not faster in a number of apps; last time I looked (or, indeed, used a mechanical drive as a system disc) SSD's were the performance king by a nautical mile. I'm not aware of any type or number of hard discs that can compete on performance with even a mid-range SSD these days.
In the 90's your only option other than an array the size of a fridge was an enterprise ramdisc (and try getting either of those into your laptop). Depending on which metric is most important to you (and it should be random read/write), SSD's are pretty much an order of magnitude faster that mechanicals.
On top of that, not everyone needs the space. Corporate laptop users especially will boot off a small 30-60GB SSD and have all their documents presented over the network or over VPN. Paying $0.10 a GB is a loss when it's cheaper for you to pay $5 a GB and get $30 more productivity from an employee per day due to faster boots, shorter logins, less thrashing during the mandatory virus scan and fewer mechanical failures (which is often enough to put an employee out of action for a significant fraction of the day, or even week if they're out and about).
IIRC Intel (who make up 2 out of 5 of the SSD's I currently own) say you can write something like 80GB to one of their 160GB drives every day and not start running out of sectors for five years; I don't know of any mechanical discs (especially 2.5" SATA ones) that are capable of such punishment. Flash and SSD controllers (along with a DRAM cache and NCQ) have really come a long way in mitigating the rewrite/longevity problem.
Disclaimer: I don't understand why more people don't love SSD's, especially when the dislike seems to relate solely to their high price.
In my experience, most of the reason the "loading" bar is there all the time is because of all the slow-arsed ad and JS servers dotted all over the place, with 90% of web pages. I use Opera mostly myself, and it'll generally tell you which sites it's waiting on when loading a page - which is often handy for adding entire domains to opera's global block list.
I've ballpark tested this myself - I yanked down a copy of several sites in httrack and loaded them on the slowest web server I have access to (running on an ARM in a friends NAS), and then limited the speed of the ethernet port to 1Mb. Even when thrashing the wireless in order to add as much latency to the network as possible, the pages were still rendered virtually instantly, as all those external references also loaded almost instantaneously.
If I paid attention to driving in the movies, you need to constantly sway the steering left and right in order to stay in a straight line, and every time you declutch and shift a gear you need to do an elaborate jump-cut to a close up of your foot on the pedal, and then your hand on the gearstick. Neither of those seem like particularly safe practices to me.
Still, at least I've taken their advice about caravans to heart - those things are death traps! http://www.youtube.com/watch?v=aaiA9ksZGS0 (5:30)
In definitely-not-related-news; Novell are in serious financial trouble, and hence the possibility of losing the faith-based patent amnesty/MAD Microsoft have with Novell.
Everyone else bitching about Mono has known about this possibility for years.
If I hadn't already spammed this thread with posts I'd have modded you up - other than price/GB, most of the problems in the wiki article are exaggerated IMHO.
I don't get the issue with the difference between read and write speeds though. Hard drives are slower ar writing too, usually about 2/3 the speed of a read for sequential access - all the SSD's I've used have a similar discrepancy (thanks to clever controllers and TRIM making writes far faster).
As to the longevity issue, I guess the jury is still out on that as these things have only been in the wild for a couple of years. But IIRC Intel said you could write 1/3 of the capacity to their drives every day for five years before you start to run out of writable blocks, and even then it's meant to fail gracefully.
I don't understand the point about SATA-based SSD's being slower for writes either; even the crappiest SSD I've used (an OCZ Agility) has random 4k writes somewhere in the order of 8-10 times faster than a 10k SAS drive. Smells like baldercrap to me but I'll go give the sources a read - here's looking forward to your edit!
Most likely as a new sort of controller hanging off the PCIe bus with a generic interface for plugging in flash cards of X gigabytes. Most of the ATA protocol is geared around spinning platters, and as such has a whole load of folderol that it and the OS needn't bother with for solid state media. KISS!
And in the event your post was intended as a jest, I'd rather see it implemented as a softraid driver of USB floppies, each connected to a 1.44MB flash chip ;)
Figured you were a programmer (I'm an infrastructure guy so have only a murky view of the internal workings of software) - part of the question about the apps was why two different speeds of memory would be an issue - I figured it'd be bad practice for an app to assume that it'd always be able to map memory at 20GB/s or whatever, but I don't really know the ins and outs of low(ish)-level programming.
I figured it'd work in a similar way to file cache now - a map of blocks/tables is given to the OS to use for storage, and the kernel just alternates between flagging these blocks as in or out of memory as applications demand - effectively using memory as read/write-through cache. As soon as a DMA came in, the OS would read those blocks from SSD into memory transparently, and keep them there for as long as necessary. Similarly, a file object that's opened read/write in memory and has been static for X seconds would be transparently written back to the SSD, or immediately written back to disc as soon as the file handle is closed. I figure a small battery and/or capacitors would be able to handle keeping the data cached (or even written) in the event of sudden power loss to help prevent corruption, plus I'm pretty sure you could implement journaling as well.
But again, IANAP so please feel free to pick holes in my naive assumptions :)
Agreed, and the nearline storage we implemented in my response to amorsen's post was along those lines - the "software" we wrote was about 250 lines of python and a couple of bash scripts for maintenance, with the rather awesome rsync+hardlinks doing the heavy lifting - I use a similar homebrew thing for backing up my home network, which implements (whole file) "deltas", date/time versioning and date pruning. I'd love to do the same thing at work with access to "proper" backup hardware + APIs (for things like taking consistent snapshots of DBs without dumping them to file) but the money for it isn't there.
Point is very much taken about tape reliability (although hard drives really don't travel well at all), we've had at least two that came back corrupt following a test DR. However, most of our contracts mandate tape in offsite escrow, and no beancounter wants to pay for both hard drive AND tape backup.
Indeed, yes, you're correct, alot of this is all about the terminology - I tend to conflate the terms myself, but then the distinction between the two can be very murky. True, most of the time a backup of, say, a commonly used database will be next to useless of it's more than a month or two old, but I've run into scenarios where we've had silent corruption in a (thankfully infrequently used) DB that the DBAs didn't notice for nearly six months, which resulted in us pulling the archival tape and restoring from that.
Personally, I wish we had the money for a nearline HDD backup or a VTL in my current job, but seeing as we *have* to have tape for archival reasons and we can turn most restores around fairly quickly, it's not seen as a necessary expense.
Dunno about magentic leakage as tapes aren't really my area (mainly because I hate the damned things :))
Personally, I don't really think people using a single tape drive without a robot/library count as "enterprisey" enough for my blood - if I had a single drive, I too would use a nearline HDD-based storage system to keep the last few versions close at hand so there's no need for the rigamarole of spooling through a tape to restore just one file. We implmented just such a "on the cheap" system in a previous job some years ago - we backed up to tape directly from the file server, but also maintained a backup server that rsynced the files over every hour. It used hardlinks to allow us to keep several faux-snapshots of the entire system whilst only actually keeping one instance of every unique file, giving us about 4 months of instant restore-any-file-you-like in just 2x the amount of space in use on the file server. Easy to set up and saved us a fortune on backup infrastructure.
I believe mainframes already do the whole memory page == disc block thing, but I'm not an expert.
I asked the same question on /. a while back regarding more common operating systems and got this response from m.dillon, which seems to indicate it's not really feasible unless the whole software stack is (very) radically altered, or the performance delta between memory and storage becomes alot smaller.
It was probably a rhetorical question, but I'll answer anyway: lots. Lots and lots and lots.
We've in the middle of replacing 48U's worth of short-stroked fibre channel discs with 4U's worth of solid state drives. Capacity was never much of an issue with these databases (they only total about 800GB) but to get the performance with an IBM pSeries box cost stupid money - I don't know the exact figure but it was somewhere in the region of 50k a year just for maintenance.
Even if you're not using hardware like a pSeries, a SAN or disc array capable of sustaining >20,000 IOPS is still going to cost you silly money, take up alot more space and eat at least five times more power.
I reckon we'll need a couple more years of SSD acceptance before this becomes mainstream - for one thing, we'll need an OS-agnostic method of using PCIe cards as bootable block devices, which will probably take a while to work out. Heck, I'm not even sure if FusionIO is bootable yet (I think they're working on it). And then there's all those filesystems in use that all assume they're on spinning discs.
SSD will be made of awesome when this happens though - SATA has been a bottleneck for flash for quite a while now, and removing a) the slow-assed bus and b) the complex disc controller overhead and the futzing around that the OS does will give gargantuan speed improvements. I expect there'll be a shift like there was from MFM to IDE - move the control of the hardware away from the OS and into a smart hardware controller that embedded with the device. SATA is just used now because it's common and universally supported by pretty much everything - crucial at the "early adopter" stage.
Not really - whilst spinning platters are awesome for nearline storage, they're not nearly reliable enough to be used as a backup solution IMHO. Stick a bunch of caddied HDD's in a cupboard and leave them for a year and you can practically guarantee one of them will fail to spin up. Tapes will remain in the backup for the foreseeable future because, TTBOMK, there's no other media that's reliable enough over the long term.
Big-assed RAID arrays used for backup storage are all well and good, especially in a small or low-budget environment, but enterprises use tapes for a reason.