When you are spinning a couple hundred TB of disk that is a *LOT* of money
Which is why file systems like GPFS are so cool, with their policy based allocation to different classes of disk pools, so I can write a policy like to force all my MP3's onto SATA for example like this
/* force MP3's and the like onto SATA storage forever */ RULE 'mp3' SET POOL 'sata'
WHERE LOWER(NAME) LIKE '%.mp3' OR LOWER(NAME) LIKE '%.m4a'
Or I can migrate files that have not been accessed for a some time of the fast disk onto the slow disk
/* migrate old files to SATA to keep the FC disks free */ RULE 'mig' MIGRATE FROM POOL 'system'
THRESHOLD(90,70) WEIGHT(weighting) TO POOL 'sata'
where the weighting factor is a marco that depends on last time accessed and file size. I can also reserve specific disks for metadata, and allocate those to the fastest disks for I/O I have.
I tell you all those ZFS fan boys have not the faintest clue how crap ZFS actually is for an allegedly all singing and dancing modern FS.
It seems to be a feature of MacOS users, but they do put the wackiest shit in their filenames. There is the whole putting a space at the start of the file name so it comes out top in finder shit, and I I thought I has seen it all with question marks, asterix's and back ticks.
How wrong was I though, because they put newlines in the middle of a file name. You have got to be trying some to manage that one, and it was not random accident because I had thousands of the files like this.
But it does not do quota's so that is not much use then in a real working environment. I can hardly build a SMB server to hand out profiles to a few thousand users if I cannot quota it.
It is also not clustered, so I cannot deploy a high availability, high throughput clustered Samba solution using CTDB. This is sucking quite a bit more then.
It has no notion of storage pool hierarchy, so I cannot mix 15kr pm SAS/FC with 7.2k rpm SATA and write policies to allocate different files to different disks, and move files between the disk types. Hum, it is now sucking quite a lot then.
It also has no HSM, so actually pretty useless really, and I can go back to using IBM's GPFS which is a *vastly* superior file system.
Lets start adding lead back into our fuel then, because that increases the mileage. Or perhaps we should take out all those side impact bars and other safety features that increase the mass of the car, as getting rid of them would improve them mileage as well.
Clearly one should care how your car achieves the mileage it does.
The two are inextricably linked. The amount of energy in a given volume of petroleum or diesel is the constant, and extracting that energy is going to produce various emissions. Therefore if I lower the amount of CO2 I produce per mile or km traveled then I must increase the efficiency of my engine to do so, which by extension will increase the fuel economy.
The difference is whether your actions effect me? You are only free to do what you want so long as it does not negatively impact me.
As driving around in a Hummer for example *DOES* effect me, if the majority of people say you don't get to do that then you don't get to do that. It is no more nanny state than making murder a crime.
As for telling kids to pray in school, well whether they pray or not has no impact on me so the state does not get to say they must.
Sorry, but spoken word at 48kbps sounds bloody awful. You try listening to a 12 hour audio book at that sort of bit rate and it is actually quite difficult to follow and requires significant concentration.
In my personal experience (and I have thousands of hours of spoken word material), drop bellow 128kbps mono and you will notice. It is not just me either, both my brother and sister who are also audio book fans can tell the difference as well.
Did they upgrade their SAN or their storage arrays?
It is hard to see how upgrading the firmware on a bunch of fibre channel switches could hose the data on the disk.
On the other hand I can think of plenty of ways a firmware upgrade to a storage array can hose all your data. The first one is *NEVER* upgrade more than one shelf at a time. The second is *NEVER* upgrade the firmware in a more than one hard disk in a RAID array at a time. Don't believe *ANY* vendor that tells you it is fine to select all the shelves etc. in your array any upgrade them in a single go. If your storage array does not allow one shelf etc. at a time, when you replace it pick a vendor that does allow this.
That's bollocks. I routinely backup a 70TB file system, where routinely is every dam day. You just need to get better tools. For the record I use IBM's TSM for the backup and GPFS for the file system.
Apart from being a potential nifty space drive, it would also provide a new test of General Relativity. This is far more likely to get it done as a real experiment at the LHC, than a new space drive.
Because 200mph+ high speed rail is a proven technology, with billions of passenger journeys already undertaken. On the other hand maglev is basically an experimental technology. I would also point out that the TGV has been experimentally run at 357mph, though to do this for an operational service would require more development.
Or an Audi A2, which has the same coefficient of drag. Or perhaps a Mercedes E-class which is lower at 0.24, or perhaps the Tatra T77 from way back in 1935 which at 0.212 has the lowest coefficient of drag of any production vehicle ever. The EV1 was lower still at 0.195
The Prius is a stupid bit of engineering pandering to the US's obsession with gasoline engines as cars such as the Audi A2 in the 1.2 and 1.4 TDi and the Polo and Golf Bluemotion all of which have better mileage figures and lower CO2 emissions.
Well George Michael is a convicted criminal so I won't bother listening to what he has to say with regards the law.
Lilly Allen has engaged in blatant copyright ingredient, on the web including infringing mix tapes, and copying of newspaper and other articles. Let he with out sin throw the first stone...
However it is heavily USA biased, so does not necessarily tell us anything. For example if you looked at Premiership footballers born in England you would find that they are heavily biases to people born between the months of September and December.
The explanation being that school years start in September and the oldest tend to be the biggest and strongest, so are better at physical activity that those that could be almost a year younger, and while they eventually catch up it is too late.
Talk to any early years teacher an they will tell you that the oldest in the year tend to do better than those that again could be almost a year younger. Do they ever catch up, can they ever catch up?
I call bullshit. Coming out of a hibernation state is *much* faster than a boot, because all your applications are open and ready to go exactly where you left off.
My laptop has 4GB of RAM and a single 7200rpm drive and I can assure you it is much better hibernating than booting.
In addition I can change batteries without loosing suspend, so you could do to get a better laptop if yours does not let you do that.
No the claim is that moving genes from unrelated species can only be done using direct cellular genetic engineering techniques, which makes "GMO's" (ignoring the fact that all food we eat is genetically modified) fundamentally different from selective breading.
The point is that this notion is not only completely unsupported by any evidence, but the actual scientific evidence is that is that unrelated species can and indeed do swap genes.
The reality is that people who argue against genetically modified food do so from a complete lack of understanding of the fundamental science involved. When you do understand the science they just look like a bunch of ignorant crazed loons.
Install GPFS and then throw TSM into the mix and you have just that today. Some fast 15k spindles, tiered to some slow but large 7.2K SATA spindles, tiered finally to a large tape library.
Then throw in CTDB for some clustered Samba and node resiliency.
Given that I can buy a storage controller that is able to do RAID 6 reads and writes with *NO* performance penalty what is this rubbish about CPU usage for? That most storage vendors are lame and don't do such controllers at the moment is not really relevant.
You have equated cost to overall CO2 production, and environmental degradation. This is faulty thinking, I have never claimed it is not cheaper to use CF.
Personally I would make it mandatory that all houses sold had a minimum level of insulation. In addition I would offer interest free loans to repaid on sale for insulation. Given the average occupancy of a home is a only seven years it should make a real difference relatively quickly.
Had that been done 15 years ago we would be sitting pretty now.
Slashdot Mirror
When you are spinning a couple hundred TB of disk that is a *LOT* of money
Which is why file systems like GPFS are so cool, with their policy based allocation to different classes of disk pools, so I can write a policy like to force all my MP3's onto SATA for example like this
RULE 'mp3' SET POOL 'sata'
WHERE LOWER(NAME) LIKE '%.mp3' OR LOWER(NAME) LIKE '%.m4a'
Or I can migrate files that have not been accessed for a some time of the fast disk onto the slow disk
RULE 'mig' MIGRATE FROM POOL 'system'
THRESHOLD(90,70) WEIGHT(weighting) TO POOL 'sata'
where the weighting factor is a marco that depends on last time accessed and file size. I can also reserve specific disks for metadata, and allocate those to the fastest disks for I/O I have.
I tell you all those ZFS fan boys have not the faintest clue how crap ZFS actually is for an allegedly all singing and dancing modern FS.
Does it support random shit in the file name?
It seems to be a feature of MacOS users, but they do put the wackiest shit in their filenames. There is the whole putting a space at the start of the file name so it comes out top in finder shit, and I I thought I has seen it all with question marks, asterix's and back ticks.
How wrong was I though, because they put newlines in the middle of a file name. You have got to be trying some to manage that one, and it was not random accident because I had thousands of the files like this.
But it does not do quota's so that is not much use then in a real working environment. I can hardly build a SMB server to hand out profiles to a few thousand users if I cannot quota it.
It is also not clustered, so I cannot deploy a high availability, high throughput clustered Samba solution using CTDB. This is sucking quite a bit more then.
It has no notion of storage pool hierarchy, so I cannot mix 15kr pm SAS/FC with 7.2k rpm SATA and write policies to allocate different files to different disks, and move files between the disk types. Hum, it is now sucking quite a lot then.
It also has no HSM, so actually pretty useless really, and I can go back to using IBM's GPFS which is a *vastly* superior file system.
Yeah, but what if your file system is 70TB with 20 million files?
None, but the AC clearly does not engage in much logical thought.
Lets start adding lead back into our fuel then, because that increases the mileage. Or perhaps we should take out all those side impact bars and other safety features that increase the mass of the car, as getting rid of them would improve them mileage as well.
Clearly one should care how your car achieves the mileage it does.
The two are inextricably linked. The amount of energy in a given volume of petroleum or diesel is the constant, and extracting that energy is going to produce various emissions. Therefore if I lower the amount of CO2 I produce per mile or km traveled then I must increase the efficiency of my engine to do so, which by extension will increase the fuel economy.
The difference is whether your actions effect me? You are only free to do what you want so long as it does not negatively impact me.
As driving around in a Hummer for example *DOES* effect me, if the majority of people say you don't get to do that then you don't get to do that. It is no more nanny state than making murder a crime.
As for telling kids to pray in school, well whether they pray or not has no impact on me so the state does not get to say they must.
Get the picture?
Sorry, but spoken word at 48kbps sounds bloody awful. You try listening to a 12 hour audio book at that sort of bit rate and it is actually quite difficult to follow and requires significant concentration.
In my personal experience (and I have thousands of hours of spoken word material), drop bellow 128kbps mono and you will notice. It is not just me either, both my brother and sister who are also audio book fans can tell the difference as well.
Did they upgrade their SAN or their storage arrays?
It is hard to see how upgrading the firmware on a bunch of fibre channel switches could hose the data on the disk.
On the other hand I can think of plenty of ways a firmware upgrade to a storage array can hose all your data. The first one is *NEVER* upgrade more than one shelf at a time. The second is *NEVER* upgrade the firmware in a more than one hard disk in a RAID array at a time. Don't believe *ANY* vendor that tells you it is fine to select all the shelves etc. in your array any upgrade them in a single go. If your storage array does not allow one shelf etc. at a time, when you replace it pick a vendor that does allow this.
That's bollocks. I routinely backup a 70TB file system, where routinely is every dam day. You just need to get better tools. For the record I use IBM's TSM for the backup and GPFS for the file system.
Apart from being a potential nifty space drive, it would also provide a new test of General Relativity. This is far more likely to get it done as a real experiment at the LHC, than a new space drive.
Really? So all those Psion Organizers are just a figment of my imagination then?
Because 200mph+ high speed rail is a proven technology, with billions of passenger journeys already undertaken. On the other hand maglev is basically an experimental technology. I would also point out that the TGV has been experimentally run at 357mph, though to do this for an operational service would require more development.
Eh, far easier just to chuck something on the track mate. For an example of what can go wrong with a high speed train
http://en.wikipedia.org/wiki/Eschede_train_disaster
Though that was down to a failure of the wheel.
Or an Audi A2, which has the same coefficient of drag. Or perhaps a Mercedes E-class which is lower at 0.24, or perhaps the Tatra T77 from way back in 1935 which at 0.212 has the lowest coefficient of drag of any production vehicle ever. The EV1 was lower still at 0.195
The Prius is a stupid bit of engineering pandering to the US's obsession with gasoline engines as cars such as the Audi A2 in the 1.2 and 1.4 TDi and the Polo and Golf Bluemotion all of which have better mileage figures and lower CO2 emissions.
Well George Michael is a convicted criminal so I won't bother listening to what he has to say with regards the law.
Lilly Allen has engaged in blatant copyright ingredient, on the web including infringing mix tapes, and copying of newspaper and other articles. Let he with out sin throw the first stone...
However it is heavily USA biased, so does not necessarily tell us anything. For example if you looked at Premiership footballers born in England you would find that they are heavily biases to people born between the months of September and December.
The explanation being that school years start in September and the oldest tend to be the biggest and strongest, so are better at physical activity that those that could be almost a year younger, and while they eventually catch up it is too late.
Talk to any early years teacher an they will tell you that the oldest in the year tend to do better than those that again could be almost a year younger. Do they ever catch up, can they ever catch up?
When does thee USA school year intake start?
I call bullshit. Coming out of a hibernation state is *much* faster than a boot, because all your applications are open and ready to go exactly where you left off.
My laptop has 4GB of RAM and a single 7200rpm drive and I can assure you it is much better hibernating than booting.
In addition I can change batteries without loosing suspend, so you could do to get a better laptop if yours does not let you do that.
No the claim is that moving genes from unrelated species can only be done using direct cellular genetic engineering techniques, which makes "GMO's" (ignoring the fact that all food we eat is genetically modified) fundamentally different from selective breading.
The point is that this notion is not only completely unsupported by any evidence, but the actual scientific evidence is that is that unrelated species can and indeed do swap genes.
The reality is that people who argue against genetically modified food do so from a complete lack of understanding of the fundamental science involved. When you do understand the science they just look like a bunch of ignorant crazed loons.
Install GPFS and then throw TSM into the mix and you have just that today. Some fast 15k spindles, tiered to some slow but large 7.2K SATA spindles, tiered finally to a large tape library.
Then throw in CTDB for some clustered Samba and node resiliency.
Given that I can buy a storage controller that is able to do RAID 6 reads and writes with *NO* performance penalty what is this rubbish about CPU usage for? That most storage vendors are lame and don't do such controllers at the moment is not really relevant.
No, I am saying while it costs me more to run, it is far from clear that it causes more environmental degradation (that is CO2 and other pollutants).
Indeed I happen to live in Scotland where a significant proportion of electricity is produced is from renewable sources, mainly hydro electricity.
That further muddies the water for me. Sure if I live in Greece or southern Italy it would be different, but I don't.
The point is that a lot of effort is being expended on something that is far from clear cut in saving CO2 emissions.
You have equated cost to overall CO2 production, and environmental degradation. This is faulty thinking, I have never claimed it is not cheaper to use CF.
Personally I would make it mandatory that all houses sold had a minimum level of insulation. In addition I would offer interest free loans to repaid on sale for insulation. Given the average occupancy of a home is a only seven years it should make a real difference relatively quickly.
Had that been done 15 years ago we would be sitting pretty now.