If it won't compile clean under -Wall -ansi -pedantic, fix it.:) When I forced all the developers I worked with to always use those options, and compile clean, half their problems went away.
That was the landscape 5 or 10 years ago, recently homeschooling has gained more ground due to poor Math and Science teaching in public school, moreso than Religious reasons.
Granted, there are still those that are taking kids out of school for religious reasons, and that's generally bad because they don't give them a good science background.
"("hardware" being software on a dedicated raid controller)"
While at a very high level, this may look to be true, it's not. Hardware arrays generally used very specialized chipsets that do one thing very well, and then there is some firmware that controls everything. It's not a general program/chipset that needs to be able to do anything. Firmware + dsps, logic circuits, etc is not the same thing as Software on a general purpose CPU.
You are extremely wrong, RAID-1 should be faster on reads, but will be slower on writes if you are using a single controller. If you are mirroring across controllers, writes will be as fast as the slower drive (Though, they should always be the same exact model when you are mirroring).
Any halfway decent RAID controller, or software RAID will do large reads by using both drives at once, greatly increasing the performance.
IDE has some limits to this, because they are stupid drives compared to SCSI, but the controller can still take advantage of the 2 copies.
Uhm, it's based on the FSCK command... in the old days, when you had to use it, you were FUCKed, because you were probably going to lose data.
Therefore, FSCKd is saying the same thing... Just like FUCK used to be a law that eventually got turned into a vulgar term related to the act that was related to the law.
As someone that has well over 3000 disks spinning in my data center, SCSI definately has a much longer life-span than IDE. True, some drives will fail quickly due to manufacturing issues, but if they make it past the first 6 months, they generally last close to 4 or 5 years before the failure rates start to go up.
Compare that to the slightly more than 1000 IDE drives we use on the desktop/batch machines, which we have to swap out on a fairly regular basis.
Agreed, RAID 10 should only be used if you have at least 2 completely seperate arrays, so that the only single point of failure is the computer itself.
It's not a penalty... they just price out their Internet with the assumption you already paid for the access with Cable TV. If you don't, then they have to cover that cost.
But at least the government is here to help you...
When you have no rights to use your own brain to make decisions, don't come crying to me.
Yes, I absolutely hate this... my development has all fiber under the ground, etc... except that the company they had do it absolutely sucks (Service Electric). They don't have HD Cable, their Internet Broadband is shotty at best, and they don't offer integrated Phone billing.
We live in the Lehigh Valley, which for one reason or another, has more ISPs per-capita than most other areas, but I can't choose any of the good ones because of this stupid monopoly this crappy company has.
Uhm, the root servers are not overloaded... this has nothing to do with the root servers, this has to do with Akamai having problems.
They have a private cached network they sell access to. It's like taking a service road around crowded highways to get closer to the final destination.
One of the companies I used to work for used Akamai, nice network... not so great customer service unless you are a really big customer.
Big problem with Raid-0 is, if you lose any drive in the chain, you lose everything.
And, unless you are dealing with very large files (some games I guess) the striping size is probably too large to really help you. You are probably just putting one big data drive on there, so you aren't forcing it to stripe stuff.
RAID-1 is probably better for performance, if you are going for performance instead of size.
Depends... development servers would obviously, since that's where they develop code. But nobody except admins and production scheduling generally get shell access to production machines.
Yes, and the 011 patch also killed about 5% of the machines it was installed on before the May 4 update. Now it only kills about 1%, or about 100 machines in our case. Not to mention the several apps it killed.
SCSI is fickle? That's why it's been used primarily in production for the past 15+ years?
You could have put the drive against a same speed SCSI and IDE still would have lost, SCSI was designed to deal with the relative slowness of rotating disks.
My primary systems run on 10k rpm 147gb drives, but they outperform everything else in the shop. Reason? Multiple layers of control logic, and lots of cache:)
Slashdot Mirror
If it won't compile clean under -Wall -ansi -pedantic, fix it. :) When I forced all the developers I worked with to always use those options, and compile clean, half their problems went away.
That was the landscape 5 or 10 years ago, recently homeschooling has gained more ground due to poor Math and Science teaching in public school, moreso than Religious reasons.
Granted, there are still those that are taking kids out of school for religious reasons, and that's generally bad because they don't give them a good science background.
"("hardware" being software on a dedicated raid controller)"
While at a very high level, this may look to be true, it's not. Hardware arrays generally used very specialized chipsets that do one thing very well, and then there is some firmware that controls everything. It's not a general program/chipset that needs to be able to do anything. Firmware + dsps, logic circuits, etc is not the same thing as Software on a general purpose CPU.
You are extremely wrong, RAID-1 should be faster on reads, but will be slower on writes if you are using a single controller. If you are mirroring across controllers, writes will be as fast as the slower drive (Though, they should always be the same exact model when you are mirroring).
Any halfway decent RAID controller, or software RAID will do large reads by using both drives at once, greatly increasing the performance.
IDE has some limits to this, because they are stupid drives compared to SCSI, but the controller can still take advantage of the 2 copies.
Uhm, it's based on the FSCK command... in the old days, when you had to use it, you were FUCKed, because you were probably going to lose data.
Therefore, FSCKd is saying the same thing... Just like FUCK used to be a law that eventually got turned into a vulgar term related to the act that was related to the law.
get it?
Granted... disk still fail, or RAID wouldn't be so popular... just saying, SCSI does generally have much longer lifespans.
:)
Drives fail, just get used to it
As someone that has well over 3000 disks spinning in my data center, SCSI definately has a much longer life-span than IDE. True, some drives will fail quickly due to manufacturing issues, but if they make it past the first 6 months, they generally last close to 4 or 5 years before the failure rates start to go up.
Compare that to the slightly more than 1000 IDE drives we use on the desktop/batch machines, which we have to swap out on a fairly regular basis.
RAID-5 isn't always a clear way to go.. it does have a nasty write performance hit unless you have a very high-end controller.
Of course, for home the hit will probably be irrelavent. Just saying, it's not "THE" way to go.
Agreed, RAID 10 should only be used if you have at least 2 completely seperate arrays, so that the only single point of failure is the computer itself.
-- Keith
No need to take down, the Sun A3500 series used Raid 1+0 in it's standard config and was fully hotswappable.
I've since switched to RAID-5, but that's due to a very, very high-end array that removes the write overhead (at least to the Host).
Easy there... you would be hard pressed to find anyone besides Mormons that considers Mormons to be "Christians".
It's not a penalty... they just price out their Internet with the assumption you already paid for the access with Cable TV. If you don't, then they have to cover that cost.
But at least the government is here to help you...
When you have no rights to use your own brain to make decisions, don't come crying to me.
Yes, I absolutely hate this... my development has all fiber under the ground, etc... except that the company they had do it absolutely sucks (Service Electric). They don't have HD Cable, their Internet Broadband is shotty at best, and they don't offer integrated Phone billing.
We live in the Lehigh Valley, which for one reason or another, has more ISPs per-capita than most other areas, but I can't choose any of the good ones because of this stupid monopoly this crappy company has.
Uhm, the root servers are not overloaded... this has nothing to do with the root servers, this has to do with Akamai having problems.
They have a private cached network they sell access to. It's like taking a service road around crowded highways to get closer to the final destination.
One of the companies I used to work for used Akamai, nice network... not so great customer service unless you are a really big customer.
The Boot drive will also hold most of the primary system DLLs or Shared Objects...
It's not just about booting.
Big problem with Raid-0 is, if you lose any drive in the chain, you lose everything.
And, unless you are dealing with very large files (some games I guess) the striping size is probably too large to really help you. You are probably just putting one big data drive on there, so you aren't forcing it to stripe stuff.
RAID-1 is probably better for performance, if you are going for performance instead of size.
I guess everybody missed the sarcasm.
Most rootkits want to stay hidden... this wouldn't be the greatest way to do that :)
Depends... development servers would obviously, since that's where they develop code. But nobody except admins and production scheduling generally get shell access to production machines.
And what about when you want to replicate those settings to 50 other boxes?
IIS is one of the worst. Apache I can actually have a source-controlled config that we promote to production. IIS... not so pretty.
Administer 1000+ windows boxes, and you will realize that a GUI != a good thing.
Hence the reason there is such a big market for Patch Management, Remote administration, etc on the Windows side of the world.
GUI is great for 1 or 2 machines, after that it starts to become painful.
Do you also get automatic updates for all your programs, so that when the OS patch breaks them, they are automatically fixed?
Really, a lot of these patches flub duck something and I have to go hunt for Application updates.
Yes, and the 011 patch also killed about 5% of the machines it was installed on before the May 4 update. Now it only kills about 1%, or about 100 machines in our case. Not to mention the several apps it killed.
SCSI is fickle? That's why it's been used primarily in production for the past 15+ years?
:)
You could have put the drive against a same speed SCSI and IDE still would have lost, SCSI was designed to deal with the relative slowness of rotating disks.
My primary systems run on 10k rpm 147gb drives, but they outperform everything else in the shop. Reason? Multiple layers of control logic, and lots of cache
Actually... Scott McNealey would have to use the patents... .NET falls under most of the Java VM patents.