Except that CUPS drivers are usually very basic and with lower quality than their windows counterparts, specially with HP inkjet products. Even the HP drivers for OSX seems to be second-class when compared with Windows.
You probably mean Apple (CUPS is property of Apple). And you also forgot to mention the subpar print quality, feature-incomplete drivers and lack of accurate color management.
While at it, search for a brain on wikipedia, because it seems you're in a desperate need of one... And I doubt anyone can find your butt, lazy as you are in researching. Slide out of those two chairs and go get some air son, you'll feel better.
Actually, resilver IS a word. It refers to the act of repairing damaged mirrors and reflexive surfaces, but it is used in context as rebuild a storage unit from a replica (mirror). Also, this is the actual term used for describing the rebuild of ZFS storage units.
I'm somewhat familiarized with similar CDP systems, and every system I've seen so far has problems in specific types of failure, and most of them that support cloning/remote replication require a stable (and fast) network connection, and aren't that good recovering from incomplete replications. If you have a fire and your SAN and servers are damaged, it may take very long time reloading the vmdk files over a regular internet connection. It is an interesting option, and apparently works for you, but I ask - did you ever tried to replace the SAN and mirror it from the cloud copy?
In some countries there are legal issues in supplying personal data to 3rd parties, so a cloud copy of customer information would be illegal without proper authorization and registration. The registration requires the company to list all the people that are responsible for the data. Can you name the sysadmin of the cloud provider that has access to the copy of your storage pool?
RAID won't protect you against power surges, defective PSU, corrupted/deleted files, or that buggy update made on that essential business application. The bigger the disks, the more time it takes to rebuild the set after a failure, and the probability of another drive die because of the rebuild stress increases. Many controllers won't rebuild the dataset if another disk has a bad sector. Bigger full disks also translates to bigger chance of silent data corruption. Yes, the capacity of the available drives increases every month, but their reliability doesn't.
RAID10 with 4 disks can handle a 2 disk failure, if lucky enough, and the rebuild is less stressful and usually faster than RAID5. Also it is important to mention that raid controllers sometimes die, and reload a created volume on a different controller is less viable than expected.
RAID should be used to improve disk space, I/O speed and resiliency against disk failures, not as a backup strategy.
There usually are legal implications with that approach. I live in EU, and the law forbids the transmission of personal data to 3rd parties without the consent of the person. A company would need the client's permission to transmit the data to a 3rd party, and the customer databases should be registered with the data protection office, detailing who has access to the information. Most EU countries have somewhat similar legislation, and while it is largely ignored, sometimes a complaint may result in heavy fines.
Actually, most plastic can withstand a temperature of 80C, where printed paper may become unreadable. I wouldn't expect a tape to work after it's been heated to 80C, but I wouldn't trust a safe to be fireproof just because the manufacturer says so - for paper or plastic. There are too many environmental factors to consider.
You know, most backup software can be programmed to send SMS or emails on case of media failure, backup failure, drive failure and stuff like that. I actually prefer to use dump/restore with tapes on *nix systems (on those where the filesystem allows snapshotting) than the fancy programs, and even then is trivial to write an alert script or even automatically mail the content of/etc/dumpdates.
That only works if your data doesn't change frequently. Even if the 100Mbit link is on a local network, a 30GB incremental backup will take hours. Over the internet, the bandwidth is the obvious factor, but when using cloud providers, latency may be an issue even with high speed links.
For our web platform, we have our own backup servers (running bacula) on the same datacenter as the production servers. Using a 100Mbit connection, a full backup of one of the servers takes about 5.5h, for about 90GB of data. The incremental daily backup takes around 30m. Now imagine how much time it would take if we were backing up using a normal (consumer) internet connection. And how much time it would take to "resilver" the server from the backup, using said connection. I can probably put the double of the data (180GB) on a tape (HP Ultrium) in less than 5 hours. So you see why cloud backup is a bad idea for small business that actually have data.
1080p is one of several HD video frame formats. 1080i, for example, theoretically requires half the bandwidth, and is the most common frame format for HD broadcasting. I didn't even mentioned framerate - in cinematography, usually is 24 frames per second (1080p24), but it is common to find several framerates from 1080p23.976 to 1080p50, so saying 1080p is "tha thing" is meaningless.
HD is usually an abbreviation of HDTV, which requires that a TV must have at least one million pixels per frame, and not what you think is cool. In fact, most of HD content available today legally (except blu-ray and HD-DVD) is in fact 720p (1280x720), either unchanged, downscaled, or upscaled to 1080p.
You raise many valid points, and you are absolutely right about the removal of the serial ports from the boards. That given, usb/ethernet has "way better" error checking and a lot more bandwidth for the applications that need it. Ethernet is somewhat more resilient with long cables (though there is RS422), but probably due to cable quality requirements and the differential signal handling, and offers built-in extensibility - you can easily create and deploy your own protocols with 3rd party protocols. The failure rate you'd expect on low bandwidth serial connections isn't the same you'd expect in high bandwidth links. So if all you need is to send a couple of kb of data (such as a remote debugger console), it works fine and everything else probably is overkill. But if you need to send megabytes of information, you either roll your own CRC mechanism (which in many cases, given the ubiquity on ethernet-on-a-chip or usb-on-a-chip, translates to more complex software and the need of faster microcontrollers), or be prepared to retransmit it.
RS232 is somewhat complex to fully implement, and most RS232 devices are not real "RS232", but implement "compatible 3/5-wire" serial communication. The low bitrate speed, the one-level abstraction (the data 1's and 0's are directly translated to electrical signals), the basic error-correction and the somewhat high electrical tolerance made RS232 the cheapest way to make devices interchange data. Even in the 70's, one could troubleshoot a RS232 connection with a "cheap" oscilloscope. But with simplicity comes a price - no character encoding, no protocol auto-detection, no data compression, no "real" error correction, so comparing it to modern interfaces such as ethernet, firewire or usb is a bit like comparing apples to oranges.
You do have RS232-compatible (both electrical and mechanical) serial communications almost everywhere, specially on industrial equipment, but it is one of those technologies I won't miss (considering I've both designed serial interface systems and written software for them)
Most modern atom boards do support 4G of ram, while you're right about better performance, it comes at a price - requires active cooling (haven't seen a cheap board without it), and the cpu has a higher TDP - in fact, if you do need performance, you get a better deal (performance per watt) with an intel core i3. I agree that the GPU is somewhat crappy, but many atom applications are business-oriented and not consumer oriented. No need for divx on a POS or on a gateway. Btw, I do have a single-core atom (210 or something) and I can watch HD video without any issues - just not 1080p.
Re:silicon-on-insulator (SOI)
on
The Transistor Wars
·
· Score: 3, Informative
The histeresis of the material also applies to copper, aluminium, gold and other conductive metals using in manufacturing of the circuits.
Germanium has been used in semiconductors longer than silicon, and it is widely used today. One of the emerging alternatives to silicon is a germanium-silicon alloy, that has been gaining traction from some years now, so this is nothing new.
The Microsoft compatibility Pack for Office 2003 (that enables the opening of openxml) is actually free. I've had trouble with password protected spreadsheets, but that's it.
I actually use Thunderbird because Outlook goes apeshit with my IMAP accounts. I have a total of ~30GB of email in 4 or 5 different accounts, and with Outlook half of the times I have a unresponsive application (both 2007 and 2010). For now I'm sticking with Thunderbird, at least until they f*** it so badly I can't use it. (the interface revamp in 3.x was a joke, the new account creation wizard is absolute crap - for instance, I can't configure an email account without internet - and if this rolling release frenzy goes as well as with firefox, I'll be looking into alternatives really soon).
Your statement is a lot like asking why do anything in Java when you have 8086 assembly?
Answer: floating point. While I'm not a java guy, I've done my fair share of mathematical functions in x86 assembly (both with and without floating point support), but it is nice not to have to think about specific implementations - after you've mastered them. I'd never learned about taylor series, fixed point math, square root approximations if not for the need to implement them from zero. Its a bit like painting in watercolor - you may take years before you can paint something decent, but after you mastered the technique, you can paint anything. The use of calculators during the learning process as anything other than confirmation tools hurt the mastering of the technique. The same way that an ace in Corel Painter isn't necessarily an artist.
Slashdot Mirror
Does your shelf runs Linux?
That's why in BSD systems, you can mark a file as append-only, and with securelevel >=1 not even root can remove the flag
Except that CUPS drivers are usually very basic and with lower quality than their windows counterparts, specially with HP inkjet products. Even the HP drivers for OSX seems to be second-class when compared with Windows.
CUPS belongs to Apple since at least 2007. Check the footer on their website http://www.cups.org/
You probably mean Apple (CUPS is property of Apple). And you also forgot to mention the subpar print quality, feature-incomplete drivers and lack of accurate color management.
While at it, search for a brain on wikipedia, because it seems you're in a desperate need of one... And I doubt anyone can find your butt, lazy as you are in researching. Slide out of those two chairs and go get some air son, you'll feel better.
Try again. http://www.oed.com/view/Entry/163627?redirectedFrom=resilver#eid
Actually, resilver IS a word. It refers to the act of repairing damaged mirrors and reflexive surfaces, but it is used in context as rebuild a storage unit from a replica (mirror). Also, this is the actual term used for describing the rebuild of ZFS storage units.
I'm somewhat familiarized with similar CDP systems, and every system I've seen so far has problems in specific types of failure, and most of them that support cloning/remote replication require a stable (and fast) network connection, and aren't that good recovering from incomplete replications. If you have a fire and your SAN and servers are damaged, it may take very long time reloading the vmdk files over a regular internet connection. It is an interesting option, and apparently works for you, but I ask - did you ever tried to replace the SAN and mirror it from the cloud copy?
In some countries there are legal issues in supplying personal data to 3rd parties, so a cloud copy of customer information would be illegal without proper authorization and registration. The registration requires the company to list all the people that are responsible for the data. Can you name the sysadmin of the cloud provider that has access to the copy of your storage pool?
RAID won't protect you against power surges, defective PSU, corrupted/deleted files, or that buggy update made on that essential business application. The bigger the disks, the more time it takes to rebuild the set after a failure, and the probability of another drive die because of the rebuild stress increases. Many controllers won't rebuild the dataset if another disk has a bad sector. Bigger full disks also translates to bigger chance of silent data corruption. Yes, the capacity of the available drives increases every month, but their reliability doesn't.
RAID10 with 4 disks can handle a 2 disk failure, if lucky enough, and the rebuild is less stressful and usually faster than RAID5. Also it is important to mention that raid controllers sometimes die, and reload a created volume on a different controller is less viable than expected.
RAID should be used to improve disk space, I/O speed and resiliency against disk failures, not as a backup strategy.
There usually are legal implications with that approach. I live in EU, and the law forbids the transmission of personal data to 3rd parties without the consent of the person. A company would need the client's permission to transmit the data to a 3rd party, and the customer databases should be registered with the data protection office, detailing who has access to the information. Most EU countries have somewhat similar legislation, and while it is largely ignored, sometimes a complaint may result in heavy fines.
Actually, most plastic can withstand a temperature of 80C, where printed paper may become unreadable. I wouldn't expect a tape to work after it's been heated to 80C, but I wouldn't trust a safe to be fireproof just because the manufacturer says so - for paper or plastic. There are too many environmental factors to consider.
You know, most backup software can be programmed to send SMS or emails on case of media failure, backup failure, drive failure and stuff like that. I actually prefer to use dump/restore with tapes on *nix systems (on those where the filesystem allows snapshotting) than the fancy programs, and even then is trivial to write an alert script or even automatically mail the content of /etc/dumpdates.
That only works if your data doesn't change frequently. Even if the 100Mbit link is on a local network, a 30GB incremental backup will take hours. Over the internet, the bandwidth is the obvious factor, but when using cloud providers, latency may be an issue even with high speed links.
For our web platform, we have our own backup servers (running bacula) on the same datacenter as the production servers. Using a 100Mbit connection, a full backup of one of the servers takes about 5.5h, for about 90GB of data. The incremental daily backup takes around 30m. Now imagine how much time it would take if we were backing up using a normal (consumer) internet connection. And how much time it would take to "resilver" the server from the backup, using said connection. I can probably put the double of the data (180GB) on a tape (HP Ultrium) in less than 5 hours. So you see why cloud backup is a bad idea for small business that actually have data.
1080p is one of several HD video frame formats. 1080i, for example, theoretically requires half the bandwidth, and is the most common frame format for HD broadcasting. I didn't even mentioned framerate - in cinematography, usually is 24 frames per second (1080p24), but it is common to find several framerates from 1080p23.976 to 1080p50, so saying 1080p is "tha thing" is meaningless.
HD is usually an abbreviation of HDTV, which requires that a TV must have at least one million pixels per frame, and not what you think is cool. In fact, most of HD content available today legally (except blu-ray and HD-DVD) is in fact 720p (1280x720), either unchanged, downscaled, or upscaled to 1080p.
You raise many valid points, and you are absolutely right about the removal of the serial ports from the boards. That given, usb/ethernet has "way better" error checking and a lot more bandwidth for the applications that need it. Ethernet is somewhat more resilient with long cables (though there is RS422), but probably due to cable quality requirements and the differential signal handling, and offers built-in extensibility - you can easily create and deploy your own protocols with 3rd party protocols. The failure rate you'd expect on low bandwidth serial connections isn't the same you'd expect in high bandwidth links. So if all you need is to send a couple of kb of data (such as a remote debugger console), it works fine and everything else probably is overkill. But if you need to send megabytes of information, you either roll your own CRC mechanism (which in many cases, given the ubiquity on ethernet-on-a-chip or usb-on-a-chip, translates to more complex software and the need of faster microcontrollers), or be prepared to retransmit it.
RS232 is somewhat complex to fully implement, and most RS232 devices are not real "RS232", but implement "compatible 3/5-wire" serial communication. The low bitrate speed, the one-level abstraction (the data 1's and 0's are directly translated to electrical signals), the basic error-correction and the somewhat high electrical tolerance made RS232 the cheapest way to make devices interchange data. Even in the 70's, one could troubleshoot a RS232 connection with a "cheap" oscilloscope. But with simplicity comes a price - no character encoding, no protocol auto-detection, no data compression, no "real" error correction, so comparing it to modern interfaces such as ethernet, firewire or usb is a bit like comparing apples to oranges.
You do have RS232-compatible (both electrical and mechanical) serial communications almost everywhere, specially on industrial equipment, but it is one of those technologies I won't miss (considering I've both designed serial interface systems and written software for them)
Most modern atom boards do support 4G of ram, while you're right about better performance, it comes at a price - requires active cooling (haven't seen a cheap board without it), and the cpu has a higher TDP - in fact, if you do need performance, you get a better deal (performance per watt) with an intel core i3. I agree that the GPU is somewhat crappy, but many atom applications are business-oriented and not consumer oriented. No need for divx on a POS or on a gateway. Btw, I do have a single-core atom (210 or something) and I can watch HD video without any issues - just not 1080p.
The histeresis of the material also applies to copper, aluminium, gold and other conductive metals using in manufacturing of the circuits.
Germanium has been used in semiconductors longer than silicon, and it is widely used today. One of the emerging alternatives to silicon is a germanium-silicon alloy, that has been gaining traction from some years now, so this is nothing new.
The Microsoft compatibility Pack for Office 2003 (that enables the opening of openxml) is actually free. I've had trouble with password protected spreadsheets, but that's it.
I'd love to see your technical solution to MS Exchange that isn't IMAP with Outlook.
I actually use Thunderbird because Outlook goes apeshit with my IMAP accounts. I have a total of ~30GB of email in 4 or 5 different accounts, and with Outlook half of the times I have a unresponsive application (both 2007 and 2010). For now I'm sticking with Thunderbird, at least until they f*** it so badly I can't use it. (the interface revamp in 3.x was a joke, the new account creation wizard is absolute crap - for instance, I can't configure an email account without internet - and if this rolling release frenzy goes as well as with firefox, I'll be looking into alternatives really soon).
Your statement is a lot like asking why do anything in Java when you have 8086 assembly?
Answer: floating point. While I'm not a java guy, I've done my fair share of mathematical functions in x86 assembly (both with and without floating point support), but it is nice not to have to think about specific implementations - after you've mastered them. I'd never learned about taylor series, fixed point math, square root approximations if not for the need to implement them from zero. Its a bit like painting in watercolor - you may take years before you can paint something decent, but after you mastered the technique, you can paint anything. The use of calculators during the learning process as anything other than confirmation tools hurt the mastering of the technique. The same way that an ace in Corel Painter isn't necessarily an artist.