Frankly, that is stupid. Use whatever YOU like, regardless of whatever others like. Just because a lot of people use Ubuntu, or Knoppix or Fartardix or whatever other distro does not mean you should not try something else. Nobody is telling them to go try Gentoo here. OpenSUSE has a pretty good community and great documentation. The developers are very active and the users are pretty friendly with good information. Often times, I have seen the blind leading the blind on the Ubuntu forums, while there are also other good answers as well. That can be confusing to a new user. Your idea of using what is popular means people should all just use Windows and never try anything else.
Roy has a problem and nothing will change his mind. Apparently, nobody can have different opinions, and he has even done what I consider to be yellow journalism in this regard.
Three years ago you might have been right about it being slow. Now, you're talking a few seconds at most here and there for the difference in package management. Also, YaST is so much more than a package manager that comparing it to Synaptic is like comparing a space shuttle to a toy rocket ship.
As was also pointed out, the copying of the DVD is so that you can install without having to use the net, which not everybody has at all times
Additional info: http://en.opensuse.org/Screenshots/11.2 for some screen shots. Also, to clarify my previous post; YaST is similar, but more powerful than Microsoft's control panel. If you configure (And you should) the Packman repository (A repository is a collection of install packages that you set up by adding the URL to a window or the file directly, and you can do it easily in YaST), you should be able to isntall almost anything you want right from YaST's package management window without having to search on many web pages.
11.2 is very polished and nice and YaST makes it easy to get things done if you are unfamiliar with how Linux configuration files work or are located. It can install software almost as easily as Ubuntu, but in my opinion, does more things "right" than Ubuntu does. It has almost everything you would need and good overall integration.
If you install it, I recommend you do it via the install DVD rather than the live CD.
Not more of this shit again.... A Trojan Horse is NOT a virus. It IS malware, but a virus tends to replicate and trojan horses do not, on their own. A trojan horse is just a program is the infection (In that it does something other than wanted or specified, and does so intentionally)
I know how to use Ubuntu........ I still don't like it. I'd rather use Debian if I had to use a Debian based distro, and I don't want to have to use one.
Another person experiences the joy of OpenSUSE over Ubuntu.... It never ceases to amaze me how many people think that Ubuntu is this pinnacle of ease of use and that any other distro is like trying to type on a terminal screen in 1985.
I agree that Ubuntu gets a lot of credit it does not deserve. That's why I choose OpenSUSE as my distro.:). I realize that Ubuntu has taken on some cult status that ZOMG it is Linux, and I can use it, but lots of other distros have gotten easier to use as well. Not everything is gentoo.
Yeah....NO missed deadlines or shoddy work from outsourced companies has been reported.
In other news...Nothing like this happens from outsourcing:
We did not write the code to your specifications because it did not include pseudo-code and all the appropriate functions, diagrams and specifics that you asked for. We are glad to charge you for our effort, however. Remember, we have 3,000 coders here who have 15yrs experience and would love to work for $11US/hr......
Linux does not have a UI, the distribution chooses a desktop environment and window manager, etc to their specifications. Look at KDE 4.2 and later and not a Gnome based distribution if you want to see a nice UI.
Lots and lots of brown, and shit that spontaneously stops working. That was just my experience, your mileage may vary. There's more to it than just package management. There are other differences when going to a Debian based distro, like managing initialization and system tools. I'd recommend OpenSUSE which also releases a new version in November.
This guy wants to appear like he is technical but only makes himself look foolish. The operating system was compromised. The protections should have been put in place to limit his rights to access the system and should always have been done so under supervision and with logging. They were foolish, he did what was to be expected
I would not recommend partial stripe-set writes in software either. It can be done but defeats the purpose of RAID 10 in many ways.
Also, my comment about 100MB/s was in relation to something like each drive putting out that much on a regular basis. I doubt, very much, that it can be sustained.
Yes it is, but the issue with RAID 10, while it costs more in terms of raw monetary outlay and disk available, the benefit is that it can not only continue its performance lead during failure, but also allow easier migration of data, move half the array to another machine and be up with the full data, in a pinch. As well, RAID 5 and RAID 6 both suffer from terrible performance in degraded mode. There is a hit in performance with RAID 10, however not nearly as large.
Also, your example assumes synchronous writes on RAID 10. The great thing about it is that you can write to either disk set, as there is not a truly active/passive nature to it, and depending on the algorithm used that is just what happens. Dual writes can be written and mirrored back to each other from cache, however that is not atomic and can cause issues with coherency during a failure.
You are also only taking into account raw throughput. The overhead of having to take 14 drives and calculate the Xors for each write, to two drives, to the stripe size, will eliminate the chance of pegging the bus to each drive. As well, in their scenario, they are using multiport, which will also eliminate that advantage, though that is only in their implementation and not all should be judged by the limitations they imposed on themselves.
You can eliminate some of the performance issues with RAID 6 by adding more drives, but also, in adding drives you also increase the possibility of a failure. With RAID 10 you are roughly in the same place. As long as you do not lose both data sets you do not lose data and the rebuild process is also a lot less taxing and impacts performance a lot less.
Is this all splitting hairs? of course. At that point, I doubt a serious bottleneck is the issue. I would be more inclined to think that the fault tolerance of such a system is wonked out. Sure, they spent the money on fancy cases that do lovely things like brace the drives, but what happens to sagging power to the drives or losing power to one of the supplies. What sort of ILO do they have for management? What sort of central tool do they use to keep it all straight?
Please learn more about that which you are speaking. If you have a simle RAID 10, that is 2+2 with 1a, 1b, 2a, 2b you can lose ANY 2 drives without losing data. If you lose 1a and 1b, then 2a and 2b will have identical data to the 1a and 1b drives. Your issue is what would happen in a RAID 0+1 when a drive from both arrays has malfunctioned.
Also, in my scenario, there is no parity checksum that needs calculating. The writes are just writes and the reads can occur either on a per request or multiple request fulfillment paradigm.
You incorrectly assume that RAID 6 uses parity drives. They do not, but instead use parity striping. Every write has to include every drive. RAID 10 is faster and more efficient. As well, RAID 10 can write to either and read from either set and sync that way. You write to fewer drives with RAID 10 and read from just as many but without XOR calcs to slow you down.
The issue is not simply the throughput to the drives but the write hole parity calculation issues, as well as resync issues that can cause an outage if 2 drives happen to fail on an array with a third drive in a pending failure state. RAID 10 would offer them the ability to lose 1/2 the drives for a smaller performance penalty than losing 2 drives in a RAID 6.
Sure, you can pin the bus, and I have a 3132 in my Dell Poweredge at home that can max out around 119MB sec on 5 WD green drives. I am using RAID 10 in software (technically closer to 1E as I have 5 drives).
So, how many of you will read the article and find out they are doing software RAID.....?
As well, when dealing with 45 drives, ensuring you pull the right one from an open case is not trivial for under skilled "Ops" type people, at least with my experience having them remove the wrong cards after telling them to count 3 from the left (open) and pull the 4th card, they pull the 8th and last card....
Didn't say it was a bad choice, however they are also not saving 1 customer's data per node. RAID 10 would be faster, but at a reduced cost effectiveness.
Slashdot Mirror
Frankly, that is stupid. Use whatever YOU like, regardless of whatever others like. Just because a lot of people use Ubuntu, or Knoppix or Fartardix or whatever other distro does not mean you should not try something else. Nobody is telling them to go try Gentoo here. OpenSUSE has a pretty good community and great documentation. The developers are very active and the users are pretty friendly with good information. Often times, I have seen the blind leading the blind on the Ubuntu forums, while there are also other good answers as well. That can be confusing to a new user. Your idea of using what is popular means people should all just use Windows and never try anything else.
Roy has a problem and nothing will change his mind. Apparently, nobody can have different opinions, and he has even done what I consider to be yellow journalism in this regard.
Members include: http://www.openinventionnetwork.com/about_members.php
Three years ago you might have been right about it being slow. Now, you're talking a few seconds at most here and there for the difference in package management. Also, YaST is so much more than a package manager that comparing it to Synaptic is like comparing a space shuttle to a toy rocket ship. As was also pointed out, the copying of the DVD is so that you can install without having to use the net, which not everybody has at all times
There are lots of good documents here: http://en.opensuse.org/Additional_package_repositories
11.2 is very polished and nice and YaST makes it easy to get things done if you are unfamiliar with how Linux configuration files work or are located. It can install software almost as easily as Ubuntu, but in my opinion, does more things "right" than Ubuntu does. It has almost everything you would need and good overall integration. If you install it, I recommend you do it via the install DVD rather than the live CD.
Not more of this shit again.... A Trojan Horse is NOT a virus. It IS malware, but a virus tends to replicate and trojan horses do not, on their own. A trojan horse is just a program is the infection (In that it does something other than wanted or specified, and does so intentionally)
I know how to use Ubuntu........ I still don't like it. I'd rather use Debian if I had to use a Debian based distro, and I don't want to have to use one.
Another person experiences the joy of OpenSUSE over Ubuntu.... It never ceases to amaze me how many people think that Ubuntu is this pinnacle of ease of use and that any other distro is like trying to type on a terminal screen in 1985.
I agree that Ubuntu gets a lot of credit it does not deserve. That's why I choose OpenSUSE as my distro. :). I realize that Ubuntu has taken on some cult status that ZOMG it is Linux, and I can use it, but lots of other distros have gotten easier to use as well. Not everything is gentoo.
Yeah....NO missed deadlines or shoddy work from outsourced companies has been reported.
In other news...Nothing like this happens from outsourcing:
We did not write the code to your specifications because it did not include pseudo-code and all the appropriate functions, diagrams and specifics that you asked for. We are glad to charge you for our effort, however. Remember, we have 3,000 coders here who have 15yrs experience and would love to work for $11US/hr......
Xfce has nothing to do with Ubuntu, per se. It is a lightweight Gnome-based desktop environment
Linux does not have a UI, the distribution chooses a desktop environment and window manager, etc to their specifications. Look at KDE 4.2 and later and not a Gnome based distribution if you want to see a nice UI.
You would much rather have system-config- than the simple "yast2" for GUI and "yast" for curses?
as an FYI, I used to use Mandrake and moved TO SUSE. The integration of tools is better. YaST for ease of use is near unbeatable.
Lots and lots of brown, and shit that spontaneously stops working. That was just my experience, your mileage may vary. There's more to it than just package management. There are other differences when going to a Debian based distro, like managing initialization and system tools. I'd recommend OpenSUSE which also releases a new version in November.
This guy wants to appear like he is technical but only makes himself look foolish. The operating system was compromised. The protections should have been put in place to limit his rights to access the system and should always have been done so under supervision and with logging. They were foolish, he did what was to be expected
I would not recommend partial stripe-set writes in software either. It can be done but defeats the purpose of RAID 10 in many ways.
Also, my comment about 100MB/s was in relation to something like each drive putting out that much on a regular basis. I doubt, very much, that it can be sustained.
Yes it is, but the issue with RAID 10, while it costs more in terms of raw monetary outlay and disk available, the benefit is that it can not only continue its performance lead during failure, but also allow easier migration of data, move half the array to another machine and be up with the full data, in a pinch. As well, RAID 5 and RAID 6 both suffer from terrible performance in degraded mode. There is a hit in performance with RAID 10, however not nearly as large. Also, your example assumes synchronous writes on RAID 10. The great thing about it is that you can write to either disk set, as there is not a truly active/passive nature to it, and depending on the algorithm used that is just what happens. Dual writes can be written and mirrored back to each other from cache, however that is not atomic and can cause issues with coherency during a failure. You are also only taking into account raw throughput. The overhead of having to take 14 drives and calculate the Xors for each write, to two drives, to the stripe size, will eliminate the chance of pegging the bus to each drive. As well, in their scenario, they are using multiport, which will also eliminate that advantage, though that is only in their implementation and not all should be judged by the limitations they imposed on themselves. You can eliminate some of the performance issues with RAID 6 by adding more drives, but also, in adding drives you also increase the possibility of a failure. With RAID 10 you are roughly in the same place. As long as you do not lose both data sets you do not lose data and the rebuild process is also a lot less taxing and impacts performance a lot less. Is this all splitting hairs? of course. At that point, I doubt a serious bottleneck is the issue. I would be more inclined to think that the fault tolerance of such a system is wonked out. Sure, they spent the money on fancy cases that do lovely things like brace the drives, but what happens to sagging power to the drives or losing power to one of the supplies. What sort of ILO do they have for management? What sort of central tool do they use to keep it all straight?
Why would you care how he/she cuts pieces of paper and pictures together? We're talking about schools here.
Please learn more about that which you are speaking. If you have a simle RAID 10, that is 2+2 with 1a, 1b, 2a, 2b you can lose ANY 2 drives without losing data. If you lose 1a and 1b, then 2a and 2b will have identical data to the 1a and 1b drives. Your issue is what would happen in a RAID 0+1 when a drive from both arrays has malfunctioned.
Also, in my scenario, there is no parity checksum that needs calculating. The writes are just writes and the reads can occur either on a per request or multiple request fulfillment paradigm.
You incorrectly assume that RAID 6 uses parity drives. They do not, but instead use parity striping. Every write has to include every drive. RAID 10 is faster and more efficient. As well, RAID 10 can write to either and read from either set and sync that way. You write to fewer drives with RAID 10 and read from just as many but without XOR calcs to slow you down.
The issue is not simply the throughput to the drives but the write hole parity calculation issues, as well as resync issues that can cause an outage if 2 drives happen to fail on an array with a third drive in a pending failure state. RAID 10 would offer them the ability to lose 1/2 the drives for a smaller performance penalty than losing 2 drives in a RAID 6.
Sure, you can pin the bus, and I have a 3132 in my Dell Poweredge at home that can max out around 119MB sec on 5 WD green drives. I am using RAID 10 in software (technically closer to 1E as I have 5 drives).
So, how many of you will read the article and find out they are doing software RAID.....? As well, when dealing with 45 drives, ensuring you pull the right one from an open case is not trivial for under skilled "Ops" type people, at least with my experience having them remove the wrong cards after telling them to count 3 from the left (open) and pull the 4th card, they pull the 8th and last card....
Didn't say it was a bad choice, however they are also not saving 1 customer's data per node. RAID 10 would be faster, but at a reduced cost effectiveness.