It seems to me that the opposite is happening, cloud ready distros are choosing systemd.
Not really...they're choosing all the extras that systemd requires to be installed.
I don't think people would have a problem with systemd if it didn't replace init, cron, syslog, autofs, ntp, etc., and require you to run its version of those demons. If systemd had more separation of packages where you could use any syslog-like program that had certain features, there would be a lot less backlash.
Servers in general do need hotplug (for example, a RAID array of hot swappable hard drives)
With hardware RAID, the OS doesn't know anything about drives being added or removed from the array, and most real production servers use some sort of hardware RAID.
That being said, all the various device node managers (udev, eudev, mdev, smdev) by themselves handle hotplug just fine. The init system doesn't need to know anything about hotplug. If you want to run a particular program on hotplug, configure your device manager to do that. And, if you want to run that program using some of the features that systemd's init portion provides (CPU limits, etc.), that's fine, too...the udev rules file can just start a "system service" on hotplug. That's all the integration needed between init and hotplug.
Try a CentOS 7 minimal install, does not even have ifconfig, lspci or a bunch of other what I would consider basic stuff.
But, with systemd dragging in a bunch of packages that many would consider to be only truly useful when a GUI is installed, the actual footprint is probably larger.
On that same note, perhaps there is a "systemd way" to do what you are trying to do with ifconfig or lspci?
Because the difference in the monthly fee between a SIM only plan and a plan with a subsidised phone adds up to much more than the subsidy on the phone.
None of this applies to the US. US Carriers that subsidize phones charge the same rates for non-"pay as you go" plans regardless of how you acquired the phone. US carriers that don't subsidize phones just add the price of the phone divided by some number of months to your monthly bill.
I have just started to build out my 10Gbps backbone at home, and get over 380MB/sec on disk transfers (RAID arrays on both ends). I've got some major tuning to do, though, as I only see about 600MB/sec on the network level. I'd like to get that up to about 900MB/sec. I have switches at several places in the house that give me 1Gbps wired to most rooms. One bedroom gets wired connectivity from DirecTV cabling, and there are two WiFi access points for various portable devices.
Otherwise, I have a CentOS box serving as an iSCSI SAN (14TB) for 3 ESX hosts. I also have a separate file server (9TB), and several user workstations (desktops and laptops).
The only hardware I use that is designed to be rack mounted is the networking gear, because most rack-mount servers are too loud.
When I said "cable companies", I was referring to their ISP portions.
And those still aren't common carriers. No ISP is. That's the whole reason they are complaining about the FCC thinking of reclassifying them under Title II, which would make them common carriers.
The first episode has a hot girl meet a couple of nerdy guys who predictable run into her burly ex-boyfriend. In the next episode, said hot chick finds some reason to take a shower in their apartment, and hilarity ensues. It seemed more fanservice than geekdom.
I actually have little problem with that whole setup, as the group I hung around with in college was seriously geeky (D&D, video games, etc.), but not quite as socially awkward. And, we did have at least one girl who we had not known for more than a week borrow our shower. Like Penny, this girl was comfortable enough with people that it didn't seem weird to her. Like Leonard and Sheldon, my roommate and I didn't strike her as people who you couldn't trust in such a situation (and we didn't do anything to violate that trust).
There are many things that happened in my college days that would be considered "too unrealistic" to show up in even a sitcom, so I don't have a problem with the situations and characters on BBT. I do have problems when the show is internally inconsistent, like when Leonard and Sheldon don't know Tweety Bird's catch phrase, but Sheldon later references other Looney Tunes characters. Because Sheldon has an eidetic memory, most of the inconsistencies are concerning him, since the writers don't remember everything, and often there are different writers on different episodes. Every show needs an internal fact checker to keep the retcon factor down, but a show that's supposed to be about science should have more than one, as making those sorts of mistakes takes you out of the moment.
I have since married and become a bit more "normal", giving up my D&D and video game habits.
Personalities aside, one of the things that was an issue early on is that the guys spent so much time "goofing off" (D&D, video games, etc.), which just isn't possible once you get to the point of having to spend so much time doing the science that the show was supposed to be about.
The primary reason I have cut back on those same sorts of things is because work takes up so much time. Even with clocking less than 45 hours per week of actual work, keeping current on new technologies stretches that out to an average of 7 hours per day, every day of the year. Add in even modest sleep requirements, and I just can't spend 48 hours straight gaming, even though I still think it's a fun thing to do.
SSDs under WinXP gradually degrade in performance, because XP doesn't support SSD TRIM.
SSDs with their own toolbox app handle this problem via scheduled tasks.
The 32bit limit of 3-something GB of RAM is a bit limiting when Firefox is chewing up 500-800MB, Thunderbird is chewing up another few hundred MB, and a handful of other background tasks chewing up 40-50MB each. Moving to Win7 meant I could put in 8GB of RAM on the box, and make use of it.
I had 12GB of RAM in my XP box for 5 years (just switched to Win 7 this year). You do know that there was a 64-bit version of XP, right? And, despite what you've heard, I had no problems finding drivers for my hardware. The only reason I moved to Win7 was because I built a new machine and had learned enough about beating down the quirks of Win7 to be able to make it work well with some older apps I have to run.
The window preview as you hover over the tasks in the task bar is addictive. Being able to see thumbnails of each application window makes it easier to pick which window to bring forward (another bonus for multi-taskers).
For you, this is important, for others (like me), not so much. First, the title bar text is more than enough for me to distinguish windows from each other. Second, the preview costs memory and seems to pop up even when you don't really need it (like if there is only one window in the taskbar icon stack), which just slows down the UI. Third, I use a multiple desktop system that I have used since Windows NT, so I don't need the extra preview (the multi-desktop manager has its own), and there are also many third-party apps that add this feature to XP. Last, if you disable features like Office's "show documents in taskbar", you don't have nearly as many windows to deal with.
I also don't need the "desktop peek" feature, because I don't have 30 windows all maximized and can't see my desktop. With 11x3 virtual desktop space, I have 33 different workspaces that can each hold a few windows sized best for interaction, and memorized for later if I choose. They can be jumped to very quickly (right mouse on the window in the workspace), and I can easily move windows between workspaces when I need to.
Remote access to Windows was added as a sort of afterthought, and it still doesn't have very good support for multiuser timesharing.
Windows Terminal Services is a complete implementation of "multiuser timesharing". It's only a "server" product, so you don't generally see it unless you are an admin. Linux doesn't really separate out "server" vs. "workstation" versions, so that may be where you are confused.
On the other hand, every Windows version has support for running programs in different user security contexts at the same time, just like Linux, and because of this there are free (as in beer, at least...some also libre) add-on products that give you many of the "server" features you are used to in the POSIX environment. For example, it's trivial to add sshd functionality to Windows because the security model allows multiple users to log in at the same time.
In Windows you have to first realise that there is such a thing as X, then you have to figure out how to get it to work in Windows, etc. In effect, if you run Windows, you are faced with an uphill struggle.
I use Windows as a desktop, and Linux on servers every day, and I think you have been blinded by the fact that you only use Linux.
Anyone tasked with "installing a database" on Linux will know about X-Windows, and if they use Windows, they will also know that there are dozens of X servers available for Windows. Cygwin takes all of 15 minutes to install and configure to have an X server running. This also gives you bash, ssh, git, etc. I interact with our git repositories solely on Windows...it's no different from using Linux.
And, I use Windows as a desktop because I don't want to "tinker with GNOME/KDE/whatever". Putting a shortcut to just about anything (program, folder, host, URL, etc.) anywhere on my Start Menu is a whole lot easier than the last time I used an X desktop manager: right click on the item, choose copy as shortcut, open the Start Menu to the folder where I want the shortcut, right click and choose paste. It may now be this easy with modern X desktops, but it's been this easy with Windows for 15 years. And, don't get me started on the lack of universal clipboard under X. It's gotten better, but there are still some apps that you can't even copy and paste text between because they don't use a common clipboard interface. For graphics, it's a complete crapshoot.
On the other hand, Microsoft has no clue how to do user elevation correctly at the GUI level, despite the fact that the Windows security model is far more robust than the simple POSIX owner/group/world system. I also think MS is going down the "pretty but not functional UI" rathole fairly quickly, and if it keeps up, Linux may end up being easier to use for even casual users.
Sure, automated initial installs have been all wrapped up in little wizard-like packages. That's not the point, it's the ongoing installation and management of packages and versions and such that you have to keep up on.
I find that management of packages on Linux is far easier than Windows.
With Linux, I just apt-get upgrade or yum update and the OS and all my applications are updated to the latest version. Windows Update only does this for the OS itself (and some select Microsoft apps), so all the other apps need to either be manually updated or you need to have dozens of auto-updaters running (either as services or scheduled tasks). I know that there are some websites that give you a one-stop updater for a lot of popular software, but there's still a lot of other software that they don't support.
The primary problem for me, though, is in being forced to pay for Windows. I build my desktop machines from components so it's not an issue there. Laptops... it's an issue.)
Even though I do use Windows as my desktop, I absolutely agree that listing the price of Windows as "$0" or "included with system" is extremely misleading. Pre-built systems from companies like Dell should list the OS as a line-item price just like every other optional feature. At the very least, if Windows is the default, you should be able to subtract it for some kind of credit.
Of course, then we'd get into situations where Linux distros installed on such systems would be priced as much or more as Windows as an "installation fee".
The term to use is "common carrier". Common carrier status is something they want to keep, and if they try to control the content of information that customers use over your network, they're in danger of losing it.
Cable companies aren't common carriers. That's the problem.
If they were common carriers, they wouldn't have to respond to DMCA complaints, because they wouldn't be liable for anything they transport over their network.
1. Many people have already purchased all the pre-1999 music they want, and now only buy new music. Prior to digital, there were a lot of replacement sales of old music.
2. It is now easy to only purchase the songs you want, so people no longer have to spend $10 for two songs, which means overall revenue is down. The solution to this is for artists to create music where every track on an album is desired.
3. "Rental" options like Spotify, Pandora, etc., don't count as sales, but are widely used by many people as their only music source.
The difference here is that these are always present, but don't interfere with the content of your account. When my Android phone installed Google Video, it didn't replace another Video app, didn't become the "default" video viewer, and wasn't added anywhere except the applications menu. Unless I actively looked for it, I would not find it.
Also, every Google app can be disabled, and all you lose is the functionality of that app (although some Google apps do use functionality from other Google apps).
It works if and only if the target system is also using LSI RAID controllers.
In the business world where you don't change the underlying OS on a critical system just because you feel like it, it's pretty easy to make sure the target hardware meets the spec.
If ZFS finds a checksum error, it thinks it's running on two big drives in a stripe with no redundancy, and it will be unable to recover the lost data.
There will be yet another level of redundancy through the cluster, as each file will be available on at least two different nodes. ZFS will help in finding discrepancies that are found at the cluster level.
Why would I prefer ZFS over DIF on FC and now SAS2 which does the checksumming in a far more comprehensive manner for me and is agnositic to the file system it is on?
ZFS checksums the data in situ and will prevent bit rot, while the methods you reference are for verifying the data hasn't changed during the transfer across a wire.
3. RAID controllers are pretty much all unique and their volumes are non portable. They are also not documented well. Your drives are useless without the controller, and even recovering with a new controller of the same type is a crapshoot.
Modern LSI RAID controllers use a completely portable format. You can move the array to any other controller that supports that same level of RAID and it can be imported into the config. I have done this successfully even when the controller chip was a completely different model.
RAID controllers are limited by their on-board chips which are typically sub-GHz RISC (ARM, Intel, MIPS) processors - an external SAS RAID controller will cost you about $2-5000 extra and have a throughput of a few 100MBps and a few 100's of IOPS.
There are sub-$1000 LSI RAID controllers that have no problem providing 500MB/sec even with 10x 5400rpm drives in RAID-6. Faster drives and more spindles plus SSD cache (handled by that same LSI controller, so it's OS-agnostic) can give apparent throughput of around 150MB/sec per spindle. For your 36 spindle scenario, that would be around 5GB/sec, which is nearly double your throughput.
Right now, I'm in the process of building a cluster where each node will use two groups of hardware RAID-6 over 10 drives, and then use a zpool to combine the storage and give me all the other ZFS features (snapshots, scrub, etc.). The whole cluster will be combined using Lustre to give around 350TB of usable storage. Based on similar build-outs that our software vendor has seen, our biggest issue will be that we only have 20Gbps total network capacity per node, and under heavy load will likely be saturated.
This is the same thing people have seen with bluetooth ear pieces. They are really nice for a small percentage of the time, but not enough to capitalize for the majority of the market.
But bluetooth headsets do what they are intended for very well...they have long battery life (days at a minimum, weeks for many models) and allow you to perform any audio interaction with your phone that you might need without having to remove your phone from its storage.
A smart watch can't do audio interactions any better than the phone, and the display doesn't provide enough extra utility (texts can be listened to via text-to-speech over a bluetooth headset) until some "killer app" is written for it. In addition, most bluetooth headsets require little to no extra software on the phone, but apps that utilize a smart watch will have to have code that specifically handles the smart watch interface.
My watch is a Tissot PRC200 Automatic. Not a very expensive watch
At US$600, it's also not "cheap".
But, you do make a good point in that people who already wear watches tend to like the style they have, and won't trade that for smart watch functionality. And, that people who wouldn't mind spending $350 on a smart watch likely already have a watch at least that expensive that they won't give up.
You'd look pretty silly trying to contort yourself in order to read the time or do a google search on an ankle monitor.
I don't think I need to remind you that there are people who—as I type this—are downloading video of somebody else contorting themselves into just such a position, silly-looking or not.
So, such a device would aid in creation of "plot" for these videos.
If you're like a lot of people, you carry a backpack/computer case with you on a regular basis. Keeping your phone safely inside that bag for most circumstances would be a benefit, freeing your pockets of the burden. You could still receive/triage incoming communications while the phone was tucked away. "Nearby" for a well designed bluetooth transceiver is 30-45 feet which is enough to keep you from having to unsheathe your phone in most circumstances.
The correct solution is to put all the "phone" functionality (antenna, transmitter, etc.) in the "watch", and use the "phone" as nothing more than a remote display and computing platform. It would be tricky to create the right split (since the watch has to have some computing power), but not impossible. The second trick would be to get the battery life of the watch high enough with the added power requirements.
The current split of "watch is a peripheral" won't appeal to enough people to make true sales inroads. Sure, Apple is going to sell a lot of these just because of the Apple name, but it's still going to be just a small percentage of iPhone owners, much less smart phone owners.
Slashdot Mirror
It seems to me that the opposite is happening, cloud ready distros are choosing systemd.
Not really...they're choosing all the extras that systemd requires to be installed.
I don't think people would have a problem with systemd if it didn't replace init, cron, syslog, autofs, ntp, etc., and require you to run its version of those demons. If systemd had more separation of packages where you could use any syslog-like program that had certain features, there would be a lot less backlash.
Servers in general do need hotplug (for example, a RAID array of hot swappable hard drives)
With hardware RAID, the OS doesn't know anything about drives being added or removed from the array, and most real production servers use some sort of hardware RAID.
That being said, all the various device node managers (udev, eudev, mdev, smdev) by themselves handle hotplug just fine. The init system doesn't need to know anything about hotplug. If you want to run a particular program on hotplug, configure your device manager to do that. And, if you want to run that program using some of the features that systemd's init portion provides (CPU limits, etc.), that's fine, too...the udev rules file can just start a "system service" on hotplug. That's all the integration needed between init and hotplug.
Try a CentOS 7 minimal install, does not even have ifconfig, lspci or a bunch of other what I would consider basic stuff.
But, with systemd dragging in a bunch of packages that many would consider to be only truly useful when a GUI is installed, the actual footprint is probably larger.
On that same note, perhaps there is a "systemd way" to do what you are trying to do with ifconfig or lspci?
Because the difference in the monthly fee between a SIM only plan and a plan with a subsidised phone adds up to much more than the subsidy on the phone.
None of this applies to the US. US Carriers that subsidize phones charge the same rates for non-"pay as you go" plans regardless of how you acquired the phone. US carriers that don't subsidize phones just add the price of the phone divided by some number of months to your monthly bill.
I get a solid 110-115MB/s, but I have SSDs.
I have just started to build out my 10Gbps backbone at home, and get over 380MB/sec on disk transfers (RAID arrays on both ends). I've got some major tuning to do, though, as I only see about 600MB/sec on the network level. I'd like to get that up to about 900MB/sec. I have switches at several places in the house that give me 1Gbps wired to most rooms. One bedroom gets wired connectivity from DirecTV cabling, and there are two WiFi access points for various portable devices.
Otherwise, I have a CentOS box serving as an iSCSI SAN (14TB) for 3 ESX hosts. I also have a separate file server (9TB), and several user workstations (desktops and laptops).
The only hardware I use that is designed to be rack mounted is the networking gear, because most rack-mount servers are too loud.
When I said "cable companies", I was referring to their ISP portions.
And those still aren't common carriers. No ISP is. That's the whole reason they are complaining about the FCC thinking of reclassifying them under Title II, which would make them common carriers.
The first episode has a hot girl meet a couple of nerdy guys who predictable run into her burly ex-boyfriend. In the next episode, said hot chick finds some reason to take a shower in their apartment, and hilarity ensues. It seemed more fanservice than geekdom.
I actually have little problem with that whole setup, as the group I hung around with in college was seriously geeky (D&D, video games, etc.), but not quite as socially awkward. And, we did have at least one girl who we had not known for more than a week borrow our shower. Like Penny, this girl was comfortable enough with people that it didn't seem weird to her. Like Leonard and Sheldon, my roommate and I didn't strike her as people who you couldn't trust in such a situation (and we didn't do anything to violate that trust).
There are many things that happened in my college days that would be considered "too unrealistic" to show up in even a sitcom, so I don't have a problem with the situations and characters on BBT. I do have problems when the show is internally inconsistent, like when Leonard and Sheldon don't know Tweety Bird's catch phrase, but Sheldon later references other Looney Tunes characters. Because Sheldon has an eidetic memory, most of the inconsistencies are concerning him, since the writers don't remember everything, and often there are different writers on different episodes. Every show needs an internal fact checker to keep the retcon factor down, but a show that's supposed to be about science should have more than one, as making those sorts of mistakes takes you out of the moment.
I have since married and become a bit more "normal", giving up my D&D and video game habits.
Personalities aside, one of the things that was an issue early on is that the guys spent so much time "goofing off" (D&D, video games, etc.), which just isn't possible once you get to the point of having to spend so much time doing the science that the show was supposed to be about.
The primary reason I have cut back on those same sorts of things is because work takes up so much time. Even with clocking less than 45 hours per week of actual work, keeping current on new technologies stretches that out to an average of 7 hours per day, every day of the year. Add in even modest sleep requirements, and I just can't spend 48 hours straight gaming, even though I still think it's a fun thing to do.
SSDs under WinXP gradually degrade in performance, because XP doesn't support SSD TRIM.
SSDs with their own toolbox app handle this problem via scheduled tasks.
The 32bit limit of 3-something GB of RAM is a bit limiting when Firefox is chewing up 500-800MB, Thunderbird is chewing up another few hundred MB, and a handful of other background tasks chewing up 40-50MB each. Moving to Win7 meant I could put in 8GB of RAM on the box, and make use of it.
I had 12GB of RAM in my XP box for 5 years (just switched to Win 7 this year). You do know that there was a 64-bit version of XP, right? And, despite what you've heard, I had no problems finding drivers for my hardware. The only reason I moved to Win7 was because I built a new machine and had learned enough about beating down the quirks of Win7 to be able to make it work well with some older apps I have to run.
The window preview as you hover over the tasks in the task bar is addictive. Being able to see thumbnails of each application window makes it easier to pick which window to bring forward (another bonus for multi-taskers).
For you, this is important, for others (like me), not so much. First, the title bar text is more than enough for me to distinguish windows from each other. Second, the preview costs memory and seems to pop up even when you don't really need it (like if there is only one window in the taskbar icon stack), which just slows down the UI. Third, I use a multiple desktop system that I have used since Windows NT, so I don't need the extra preview (the multi-desktop manager has its own), and there are also many third-party apps that add this feature to XP. Last, if you disable features like Office's "show documents in taskbar", you don't have nearly as many windows to deal with.
I also don't need the "desktop peek" feature, because I don't have 30 windows all maximized and can't see my desktop. With 11x3 virtual desktop space, I have 33 different workspaces that can each hold a few windows sized best for interaction, and memorized for later if I choose. They can be jumped to very quickly (right mouse on the window in the workspace), and I can easily move windows between workspaces when I need to.
Remote access to Windows was added as a sort of afterthought, and it still doesn't have very good support for multiuser timesharing.
Windows Terminal Services is a complete implementation of "multiuser timesharing". It's only a "server" product, so you don't generally see it unless you are an admin. Linux doesn't really separate out "server" vs. "workstation" versions, so that may be where you are confused.
On the other hand, every Windows version has support for running programs in different user security contexts at the same time, just like Linux, and because of this there are free (as in beer, at least...some also libre) add-on products that give you many of the "server" features you are used to in the POSIX environment. For example, it's trivial to add sshd functionality to Windows because the security model allows multiple users to log in at the same time.
In Windows you have to first realise that there is such a thing as X, then you have to figure out how to get it to work in Windows, etc. In effect, if you run Windows, you are faced with an uphill struggle.
I use Windows as a desktop, and Linux on servers every day, and I think you have been blinded by the fact that you only use Linux.
Anyone tasked with "installing a database" on Linux will know about X-Windows, and if they use Windows, they will also know that there are dozens of X servers available for Windows. Cygwin takes all of 15 minutes to install and configure to have an X server running. This also gives you bash, ssh, git, etc. I interact with our git repositories solely on Windows...it's no different from using Linux.
And, I use Windows as a desktop because I don't want to "tinker with GNOME/KDE/whatever". Putting a shortcut to just about anything (program, folder, host, URL, etc.) anywhere on my Start Menu is a whole lot easier than the last time I used an X desktop manager: right click on the item, choose copy as shortcut, open the Start Menu to the folder where I want the shortcut, right click and choose paste. It may now be this easy with modern X desktops, but it's been this easy with Windows for 15 years. And, don't get me started on the lack of universal clipboard under X. It's gotten better, but there are still some apps that you can't even copy and paste text between because they don't use a common clipboard interface. For graphics, it's a complete crapshoot.
On the other hand, Microsoft has no clue how to do user elevation correctly at the GUI level, despite the fact that the Windows security model is far more robust than the simple POSIX owner/group/world system. I also think MS is going down the "pretty but not functional UI" rathole fairly quickly, and if it keeps up, Linux may end up being easier to use for even casual users.
Sure, automated initial installs have been all wrapped up in little wizard-like packages. That's not the point, it's the ongoing installation and management of packages and versions and such that you have to keep up on.
I find that management of packages on Linux is far easier than Windows.
With Linux, I just apt-get upgrade or yum update and the OS and all my applications are updated to the latest version. Windows Update only does this for the OS itself (and some select Microsoft apps), so all the other apps need to either be manually updated or you need to have dozens of auto-updaters running (either as services or scheduled tasks). I know that there are some websites that give you a one-stop updater for a lot of popular software, but there's still a lot of other software that they don't support.
The primary problem for me, though, is in being forced to pay for Windows. I build my desktop machines from components so it's not an issue there. Laptops ... it's an issue.)
Even though I do use Windows as my desktop, I absolutely agree that listing the price of Windows as "$0" or "included with system" is extremely misleading. Pre-built systems from companies like Dell should list the OS as a line-item price just like every other optional feature. At the very least, if Windows is the default, you should be able to subtract it for some kind of credit.
Of course, then we'd get into situations where Linux distros installed on such systems would be priced as much or more as Windows as an "installation fee".
The term to use is "common carrier". Common carrier status is something they want to keep, and if they try to control the content of information that customers use over your network, they're in danger of losing it.
Cable companies aren't common carriers. That's the problem.
If they were common carriers, they wouldn't have to respond to DMCA complaints, because they wouldn't be liable for anything they transport over their network.
From 1999 to 2009, music sales dropped about 60%.
Much of that has to do with three things:
1. Many people have already purchased all the pre-1999 music they want, and now only buy new music. Prior to digital, there were a lot of replacement sales of old music.
2. It is now easy to only purchase the songs you want, so people no longer have to spend $10 for two songs, which means overall revenue is down. The solution to this is for artists to create music where every track on an album is desired.
3. "Rental" options like Spotify, Pandora, etc., don't count as sales, but are widely used by many people as their only music source.
The difference here is that these are always present, but don't interfere with the content of your account. When my Android phone installed Google Video, it didn't replace another Video app, didn't become the "default" video viewer, and wasn't added anywhere except the applications menu. Unless I actively looked for it, I would not find it.
Also, every Google app can be disabled, and all you lose is the functionality of that app (although some Google apps do use functionality from other Google apps).
It works if and only if the target system is also using LSI RAID controllers.
In the business world where you don't change the underlying OS on a critical system just because you feel like it, it's pretty easy to make sure the target hardware meets the spec.
If ZFS finds a checksum error, it thinks it's running on two big drives in a stripe with no redundancy, and it will be unable to recover the lost data.
There will be yet another level of redundancy through the cluster, as each file will be available on at least two different nodes. ZFS will help in finding discrepancies that are found at the cluster level.
Why would I prefer ZFS over DIF on FC and now SAS2 which does the checksumming in a far more comprehensive manner for me and is agnositic to the file system it is on?
ZFS checksums the data in situ and will prevent bit rot, while the methods you reference are for verifying the data hasn't changed during the transfer across a wire.
3. RAID controllers are pretty much all unique and their volumes are non portable. They are also not documented well. Your drives are useless without the controller, and even recovering with a new controller of the same type is a crapshoot.
Modern LSI RAID controllers use a completely portable format. You can move the array to any other controller that supports that same level of RAID and it can be imported into the config. I have done this successfully even when the controller chip was a completely different model.
RAID controllers are limited by their on-board chips which are typically sub-GHz RISC (ARM, Intel, MIPS) processors - an external SAS RAID controller will cost you about $2-5000 extra and have a throughput of a few 100MBps and a few 100's of IOPS.
There are sub-$1000 LSI RAID controllers that have no problem providing 500MB/sec even with 10x 5400rpm drives in RAID-6. Faster drives and more spindles plus SSD cache (handled by that same LSI controller, so it's OS-agnostic) can give apparent throughput of around 150MB/sec per spindle. For your 36 spindle scenario, that would be around 5GB/sec, which is nearly double your throughput.
Right now, I'm in the process of building a cluster where each node will use two groups of hardware RAID-6 over 10 drives, and then use a zpool to combine the storage and give me all the other ZFS features (snapshots, scrub, etc.). The whole cluster will be combined using Lustre to give around 350TB of usable storage. Based on similar build-outs that our software vendor has seen, our biggest issue will be that we only have 20Gbps total network capacity per node, and under heavy load will likely be saturated.
This is the same thing people have seen with bluetooth ear pieces. They are really nice for a small percentage of the time, but not enough to capitalize for the majority of the market.
But bluetooth headsets do what they are intended for very well...they have long battery life (days at a minimum, weeks for many models) and allow you to perform any audio interaction with your phone that you might need without having to remove your phone from its storage.
A smart watch can't do audio interactions any better than the phone, and the display doesn't provide enough extra utility (texts can be listened to via text-to-speech over a bluetooth headset) until some "killer app" is written for it. In addition, most bluetooth headsets require little to no extra software on the phone, but apps that utilize a smart watch will have to have code that specifically handles the smart watch interface.
My watch is a Tissot PRC200 Automatic. Not a very expensive watch
At US$600, it's also not "cheap".
But, you do make a good point in that people who already wear watches tend to like the style they have, and won't trade that for smart watch functionality. And, that people who wouldn't mind spending $350 on a smart watch likely already have a watch at least that expensive that they won't give up.
You'd look pretty silly trying to contort yourself in order to read the time or do a google search on an ankle monitor.
I don't think I need to remind you that there are people who—as I type this—are downloading video of somebody else contorting themselves into just such a position, silly-looking or not.
So, such a device would aid in creation of "plot" for these videos.
If you're like a lot of people, you carry a backpack/computer case with you on a regular basis. Keeping your phone safely inside that bag for most circumstances would be a benefit, freeing your pockets of the burden. You could still receive/triage incoming communications while the phone was tucked away. "Nearby" for a well designed bluetooth transceiver is 30-45 feet which is enough to keep you from having to unsheathe your phone in most circumstances.
The correct solution is to put all the "phone" functionality (antenna, transmitter, etc.) in the "watch", and use the "phone" as nothing more than a remote display and computing platform. It would be tricky to create the right split (since the watch has to have some computing power), but not impossible. The second trick would be to get the battery life of the watch high enough with the added power requirements.
The current split of "watch is a peripheral" won't appeal to enough people to make true sales inroads. Sure, Apple is going to sell a lot of these just because of the Apple name, but it's still going to be just a small percentage of iPhone owners, much less smart phone owners.