If I'm not mistaken, WiFi has also traditionally been half duplex. This may have changed with some of the newer N stuff tho, I haven't really kept up. So that would definitely cause problems. That being said, if a half duplex link can kill NFS performance, I can only imagine what that would have been like in the days of thinnet.
Not to mention, the only actual MAC address that would remotely matter would be the Modem's MAC address. And that would only matter until the next hop when the packet was out of that collision domain (Ok, probably more like outside the hands of your ISP since they have your information on file)
I don't see why people are so paranoid about their MAC addresses, I could maybe see in a coffee shop where someone can pull the AP's logs, but across the Internet, the only one who's gonna be able to see anything relating to your MAC address is going to be your ISP, and considering that they frequently identify customer's by the MAC address of their modem, that's not like its something that can change.
Its not like I've ever ranted about this before or anything.
Last time I checked, yeah, Amazon does have incredible infrastructure, I mean, when they go down, half the internet goes down. I can only imagine that they eat their own AWS dog food.
It seems to be a regional thing. I've heard of people getting fantastic results from Comcast, while CenturyLink in my home town is so shoddy, you're almost better off using something like a MiFi (And I hate using those with a burning passion).
I just bought a new car a few months ago, and I've definitely noticed my driving style is entirely different now.
I'm all about a simple dash, a stick shift, and few distractions; driving is one of the few times that I can sit down and focus non stop on something. In my new car, I find my self having to fight bad habits of fiddling with the radio and all the extra gizmos my car has.
Of course there comes the awkward phase of solo hobbyist transitioning to professional work. I've been studying to get into Linux administration, and I'll get talking with some of my other linux-y friends about our latest projects and there's a definite breakdown. They're talking about playing with the latest distros, I'm talking about building a storage pool for a KVM cluster or something. we just give each other mutual blank stares and change the conversation.
I've become a bit notorious for walking around the office wearing a pair of giant studio monitors plugged into my ipad. Its usually just because I don't want to dedicate another device to playing music, but I don't want to stop my music when I get up... that being said, I don't usually do it outside of work
Another suggestion might be the x120. I believe it was actually designed with the education market in mind. I'm a huge fan of their x series, and if I was to buy a new laptop today, that is probably what I would buy.
The way I see it is that there is definitely a happy medium to having a facebook. I mean, you don't have to use all the functionality of it. I don't 'like' random things I find on there play any of the games or anything, I rarely post status updates, I periodically upload a picture or 2. Its mainly just an easy central point of contact for people.
I see facebook as being similar to having my name in the phone book.
And if someone is paranoid about being tracked, well, that's what noscript and adblock+ are for.
I know the feeling. Any time I've had a tech blame it on my router, I kindly point out that my router is successfully routing traffic between every other interface... And yet they still insist on walking me through basic troubleshooting steps.
Eventually, I just flat out ask them to try reprovisioning the line instead of doubting the functionality of the devices on my end. 99% of the time that works.
Its not a matter of bad controller design, its a matter of expected behavior. The problem is that the consumer drive doesn't necessarily immediately acknowledge there's a problem. And the controller has no idea there's a problem because the disk didn't immediately acknowledge the error, so it doesn't attempt to pull the data from the other volumes until the timeout is reached. And as far as the controller is concerned, the disk is taking longer than normal to respond to a command, and thus is likely failing.
You can adjust the amount of time it waits for a response to account for consumer disks, But this degrades performance because the controller is just waiting on data that it could have gotten from a faster source.
Essentially it boils down to: Enterprise drives expect to have a fallback, and thus they will immediately report an error. Consumer drives, on the other hand, expect to not have anything to fall back on or any time limits on how long they have to respond, so they won't say there's anything wrong until they've tried to fix it.
Hypothetically, since this is primarily a matter of error reporting we're discussing, the entire problem could be solved by simply flashing raid optimised firmware onto a consumer drive. This wouldn't magically turn them into enterprise drives, but it would solve the potential problem with using them in raid.
Everyone seems to be missing the fact that most of these enterprise drives are little 2.5 inch 15k rpm speed daemons.
And as for your comment about consumer drives in raid. While it might work most of the time, the firmware that comes on the disk's controller isn't necissarily optimised for a raid environment. Minor errors pop up during everyday operations, and a normal single drive will spend some time calculating to recover from the error. This is perfect for a single drive scenerio, but If this happens in a raid environment, the controller times out and assumes the disk has gone offline. Even if the disk hasn't actually failed, This puts the volume in degraded mode until you can bring that disk back online. Disks optimized for a raid enviornment just mark the sector as bad, tell the controller, and let the other volumes pick up the slack for that piece of data-- no harm, no foul. Sure you can adjust the timeout on the raid controller, but why wait for the single disk to do its business when the raid card can just direct the task to the other disks. Minor details such as that make for a world of difference.
I have a quasi similar story to this.
So one of my friends used to have this bad habbit of saying his passwords while he was typing them, which usually wasn't a problem, but he happened to do it one time while I was on the phone with him, and it happened to be his root passqord... Whoops!
Armed with this knowledge, I also realized that he had a subdomain on his site set up to re direct to his home machine so he could access stuff from school. Soo a couple days later, we were chatting online and I feel like messing with him so I went to eject his cdrom a couple times to mess with him. Unknown to me, I accidentally ejected his zip drive instead.
The best part was, his desktop was set up so the drive was above his monitor on a shelf, so when I ejected it, the zip disk ended up hitting him right on the head.
Its not an olpc, but my EEE makes a handy platform for launching an attack, its small, its handy and it has that, its not possibly big enough to be a threat factor. Although,when I pair it with my 500 gig external drive that is happily chugging away, it probably looses a few innocence points. But overall, it is a great little platform and was worth every penny.
Just because one is an advanced student doesn't mean they will get the best teacher. An example of this my freashmen year I had a first time english teacher for my ADVANCED English... to say the least, the only people who thought she was a good teacher were the cheerleaders.
Slashdot Mirror
If I'm not mistaken, WiFi has also traditionally been half duplex. This may have changed with some of the newer N stuff tho, I haven't really kept up. So that would definitely cause problems. That being said, if a half duplex link can kill NFS performance, I can only imagine what that would have been like in the days of thinnet.
Not to mention, the only actual MAC address that would remotely matter would be the Modem's MAC address. And that would only matter until the next hop when the packet was out of that collision domain (Ok, probably more like outside the hands of your ISP since they have your information on file)
I don't see why people are so paranoid about their MAC addresses, I could maybe see in a coffee shop where someone can pull the AP's logs, but across the Internet, the only one who's gonna be able to see anything relating to your MAC address is going to be your ISP, and considering that they frequently identify customer's by the MAC address of their modem, that's not like its something that can change.
Its not like I've ever ranted about this before or anything.
I don't wear a suit, as much as I wear whatever happens to be closest to me when I wake up, but I care.
That being said, I work at HP, and its nice to know what's going on with the company so I can plan for any future Resume Generating Events.
Last time I checked, yeah, Amazon does have incredible infrastructure, I mean, when they go down, half the internet goes down. I can only imagine that they eat their own AWS dog food.
It seems to be a regional thing. I've heard of people getting fantastic results from Comcast, while CenturyLink in my home town is so shoddy, you're almost better off using something like a MiFi (And I hate using those with a burning passion).
Not even U-Haul?
You mean:
Because Racecar.
I just bought a new car a few months ago, and I've definitely noticed my driving style is entirely different now.
I'm all about a simple dash, a stick shift, and few distractions; driving is one of the few times that I can sit down and focus non stop on something.
In my new car, I find my self having to fight bad habits of fiddling with the radio and all the extra gizmos my car has.
Of course there comes the awkward phase of solo hobbyist transitioning to professional work. I've been studying to get into Linux administration, and I'll get talking with some of my other linux-y friends about our latest projects and there's a definite breakdown. They're talking about playing with the latest distros, I'm talking about building a storage pool for a KVM cluster or something. we just give each other mutual blank stares and change the conversation.
Virgin? Nope. I've seen rocky horror picture show thank you.
I've become a bit notorious for walking around the office wearing a pair of giant studio monitors plugged into my ipad. Its usually just because I don't want to dedicate another device to playing music, but I don't want to stop my music when I get up... that being said, I don't usually do it outside of work
Another suggestion might be the x120. I believe it was actually designed with the education market in mind. I'm a huge fan of their x series, and if I was to buy a new laptop today, that is probably what I would buy.
I guess they decided it was time to bring back a system with the halt catch fire operator
The way I see it is that there is definitely a happy medium to having a facebook. I mean, you don't have to use all the functionality of it. I don't 'like' random things I find on there play any of the games or anything, I rarely post status updates, I periodically upload a picture or 2. Its mainly just an easy central point of contact for people. I see facebook as being similar to having my name in the phone book. And if someone is paranoid about being tracked, well, that's what noscript and adblock+ are for.
You don't have to tell me twice. Or even once for that matter.
I know the feeling. Any time I've had a tech blame it on my router, I kindly point out that my router is successfully routing traffic between every other interface... And yet they still insist on walking me through basic troubleshooting steps. Eventually, I just flat out ask them to try reprovisioning the line instead of doubting the functionality of the devices on my end. 99% of the time that works.
Indeed. Me and my friends' corner of the action at Quake Con smelled more like Dr. Pepper and twizzlers than it did sweat.
Its not a matter of bad controller design, its a matter of expected behavior. The problem is that the consumer drive doesn't necessarily immediately acknowledge there's a problem. And the controller has no idea there's a problem because the disk didn't immediately acknowledge the error, so it doesn't attempt to pull the data from the other volumes until the timeout is reached. And as far as the controller is concerned, the disk is taking longer than normal to respond to a command, and thus is likely failing. You can adjust the amount of time it waits for a response to account for consumer disks, But this degrades performance because the controller is just waiting on data that it could have gotten from a faster source. Essentially it boils down to: Enterprise drives expect to have a fallback, and thus they will immediately report an error. Consumer drives, on the other hand, expect to not have anything to fall back on or any time limits on how long they have to respond, so they won't say there's anything wrong until they've tried to fix it. Hypothetically, since this is primarily a matter of error reporting we're discussing, the entire problem could be solved by simply flashing raid optimised firmware onto a consumer drive. This wouldn't magically turn them into enterprise drives, but it would solve the potential problem with using them in raid.
Everyone seems to be missing the fact that most of these enterprise drives are little 2.5 inch 15k rpm speed daemons. And as for your comment about consumer drives in raid. While it might work most of the time, the firmware that comes on the disk's controller isn't necissarily optimised for a raid environment. Minor errors pop up during everyday operations, and a normal single drive will spend some time calculating to recover from the error. This is perfect for a single drive scenerio, but If this happens in a raid environment, the controller times out and assumes the disk has gone offline. Even if the disk hasn't actually failed, This puts the volume in degraded mode until you can bring that disk back online. Disks optimized for a raid enviornment just mark the sector as bad, tell the controller, and let the other volumes pick up the slack for that piece of data-- no harm, no foul. Sure you can adjust the timeout on the raid controller, but why wait for the single disk to do its business when the raid card can just direct the task to the other disks. Minor details such as that make for a world of difference.
I have a quasi similar story to this. So one of my friends used to have this bad habbit of saying his passwords while he was typing them, which usually wasn't a problem, but he happened to do it one time while I was on the phone with him, and it happened to be his root passqord... Whoops! Armed with this knowledge, I also realized that he had a subdomain on his site set up to re direct to his home machine so he could access stuff from school. Soo a couple days later, we were chatting online and I feel like messing with him so I went to eject his cdrom a couple times to mess with him. Unknown to me, I accidentally ejected his zip drive instead. The best part was, his desktop was set up so the drive was above his monitor on a shelf, so when I ejected it, the zip disk ended up hitting him right on the head.
Its not an olpc, but my EEE makes a handy platform for launching an attack, its small, its handy and it has that, its not possibly big enough to be a threat factor. Although,when I pair it with my 500 gig external drive that is happily chugging away, it probably looses a few innocence points. But overall, it is a great little platform and was worth every penny.
Now those are some long videos
Go internet Rah Rah Rah
I wouldn't want to be involuntarily tagged, but I am fine doing it my self
Just because one is an advanced student doesn't mean they will get the best teacher. An example of this my freashmen year I had a first time english teacher for my ADVANCED English... to say the least, the only people who thought she was a good teacher were the cheerleaders.