and couldn't program it to prioritize based on which one was seen first, was closest, was apt to fall first based on speed/distance, or any one of many other possibilities. You could even place weights on them, and throw a die at the end as a tiebreaker. The rule should be interpreted as "allow the least harm," not "allow no harm."
So, you like free gifts. What's your address? I'll have my dog crap on your lawn, FOR FREE! In other words, I'll "give a shit," and make sure it's not "petty bullshit." That should make you happy.
Uh, no. Do some reading on diffserv. There are mechanisms to accommodate a range of bandwidth (assurance) and latency (expediency) needs. QoS is much more than simply reordering packets, and includes things like classification, marking, queue management (strict vs. RED/WRED vs. WFQ), policing, shaping, trust relationships, etc.
"If you have a 12mbit connection vs a 20mbit connection how is it you think the traffic magically figures that out so it can send you traffic at the correct rate for your link?"
For most ISPs, that would be traffic policing, although some may use traffic shaping. Look it up, you'll learn something new.
You clearly don't understand the difference between QoS and congestion control, or between TCP and UDP, or that some protocols cannot degrade gracefully. Congestion control in no way replaces proper QoS.
It doesn't even do that. QoS on a home router is only going to prioritize the traffic leaving the box. It's all on an equal basis "best effort" from there. So, it only protects you from yourself - you can make outbound Skype continue working when you have a bunch of torrents running. But, the quality of the incoming Skype will still suck - you have no control. The ISPs would like to give you that control (and charge you more for it), but no, that would apparently be evil.
So, the OP was being clueless. Unless QoS is bidirectional (for most types of traffic), and end-to-end, it provides little value.
If you think that's how it works, you really don't understand QoS, networking, or what the ISPs would like to do. They want to provide QoS within their networks. That would allow better support for things like realtime services (Netflix, VoIP, Pandora, etc.). They can't simply trust users to appropriately mark packets - you'd have some who simply marked everything as high priority.
And exactly how does your hypothetical user control incoming bandwidth with their "home router?"
I have no problem with preferential "fast lanes," as long as they use bandwidth above and beyond a guaranteed baseline (call it a CIR, or SLA rate, or whatever). You don't want to pay to use it, you've lost nothing. Some service provider wants to pay to send preferred traffic to you? It has no impact on your base rate, you've lost nothing (well, perhaps a tiny bit of latency to serialization delay, but no bandwidth). This assumes, of course, that the CPE-ISP link is undersubscribed.
If you're using a "home type router," my guess is you have no other internal hops. And, your ISP isn't going to pay any attention to how you mark QoS in what you send out.
So, exactly what do you expect that QoS support to do? QoS provides very little benefit unless it is end-to-end.
The most interesting trademark dispute is probably the one related to "Budweiser." "Jeep" is another, which started out in the public domain, but was then commercialized.
This is my favorite. You see, we lease this back from the company we sold it to - that way it comes under the monthly current budget and not the capital account.
In my experience, very few people use "Xerox" as a verb. I've much more often heard "make some copies on the Xerox machine" (and less often without the "machine") referring generically to a photocopier.
In any case, bad example, as Xerox still holds their trademark.
" preclusion against running 'servers' on residential service."
What's a "server?" A piece of software with a local display and keyboard connecting to the net is called a client if that piece of software is named "web browser" and a server if it is named "X windows." "Server" is an entirely arbitrary distinction.
If the NSA and federal government didn't change after the info was released publicly, why are they acting like an internal complaint might have made a difference?
In what way do you consider the choice of a measurement which is easily reproducible virtually anywhere worldwide "arbitrary?"
Yes, things like altitude change the scale a bit. Can you can come up with a better solution (very accessible, reasonably accurate, reasonably reproducible) for transfer of a standard temperature scale worldwide with mid-1700's technology? Choosing the freezing and boiling points of water on that basis for something of scientific, industrial and commercial use seems anything but "arbitrary."
"Why the hell are we talking about the Fahrenheit scale. And, while we're at it, memory of all kinds is always expressed in GiB, so a 512GB card is 1024 times as large as a 512MB card"
I use Rankine, you insensitive clod.
And, you're wrong. The Sandisk 512 GB card being discussed has a capacity of 512,000,000,000 bytes ("1GB=1,000,000,000 bytes" - Sandisk). Just like disk drives and SSDs are measured.
The discussion is about capacity, and disk drives, SSDs, and yes even these memory cards ("1 GB = 1,000,000,000 bytes" - Sandisk) use the metric prefixes correctly. How you can claim that's a "base 10 number of base 2 blocks" is a mystery. Yes, they may address base 2 sized blocks (e.g. 512 or 4096), but the total capacity is specified in base 10. (and the block address is expressed on the interface with a base 2 number, not BCD.) But that's not the capacity, which is what's being discussed.
It seems like disk manufacturers don't reveal specifics like they used to, but it wasn't uncommon to find organizations like 17 sectors/track and other non-base 2/10 layouts.
Slashdot Mirror
You're doing it wrong. The closer you are, the easier it is to hide.
and couldn't program it to prioritize based on which one was seen first, was closest, was apt to fall first based on speed/distance, or any one of many other possibilities. You could even place weights on them, and throw a die at the end as a tiebreaker. The rule should be interpreted as "allow the least harm," not "allow no harm."
So, you like free gifts. What's your address? I'll have my dog crap on your lawn, FOR FREE! In other words, I'll "give a shit," and make sure it's not "petty bullshit." That should make you happy.
"For IPv4, QoS simply means reordering packets"
Uh, no. Do some reading on diffserv. There are mechanisms to accommodate a range of bandwidth (assurance) and latency (expediency) needs. QoS is much more than simply reordering packets, and includes things like classification, marking, queue management (strict vs. RED/WRED vs. WFQ), policing, shaping, trust relationships, etc.
"If you have a 12mbit connection vs a 20mbit connection how is it you think the traffic magically figures that out so it can send you traffic at the correct rate for your link?"
For most ISPs, that would be traffic policing, although some may use traffic shaping. Look it up, you'll learn something new.
You clearly don't understand the difference between QoS and congestion control, or between TCP and UDP, or that some protocols cannot degrade gracefully. Congestion control in no way replaces proper QoS.
"Simple they manage the outbound rate at which they send ACKs and let TCP on the rremote host figure out the rate limiting."
That's congestion control, not QoS. Many of the protocols where QoS is most desirable run over UDP, not TCP.
Your whole "defaults to EF, demote to EF" thing is confused. I think you mean DSCPs DF, EF and AF13, where EF is the extra-cost premium service.
It doesn't even do that. QoS on a home router is only going to prioritize the traffic leaving the box. It's all on an equal basis "best effort" from there. So, it only protects you from yourself - you can make outbound Skype continue working when you have a bunch of torrents running. But, the quality of the incoming Skype will still suck - you have no control. The ISPs would like to give you that control (and charge you more for it), but no, that would apparently be evil.
So, the OP was being clueless. Unless QoS is bidirectional (for most types of traffic), and end-to-end, it provides little value.
If you think that's how it works, you really don't understand QoS, networking, or what the ISPs would like to do. They want to provide QoS within their networks. That would allow better support for things like realtime services (Netflix, VoIP, Pandora, etc.). They can't simply trust users to appropriately mark packets - you'd have some who simply marked everything as high priority.
And exactly how does your hypothetical user control incoming bandwidth with their "home router?"
I have no problem with preferential "fast lanes," as long as they use bandwidth above and beyond a guaranteed baseline (call it a CIR, or SLA rate, or whatever). You don't want to pay to use it, you've lost nothing. Some service provider wants to pay to send preferred traffic to you? It has no impact on your base rate, you've lost nothing (well, perhaps a tiny bit of latency to serialization delay, but no bandwidth). This assumes, of course, that the CPE-ISP link is undersubscribed.
"Isn't QoS supported by most home type routers,"
If you're using a "home type router," my guess is you have no other internal hops. And, your ISP isn't going to pay any attention to how you mark QoS in what you send out.
So, exactly what do you expect that QoS support to do? QoS provides very little benefit unless it is end-to-end.
Stop being ignorant. You've obviously never read Smith v. Maryland. Buh bye!
So, the 4th A isn't a legal problem. OK.
The most interesting trademark dispute is probably the one related to "Budweiser." "Jeep" is another, which started out in the public domain, but was then commercialized.
Ah, I see you have the machine that goes "Bing!"
This is my favorite. You see, we lease this back from the company we sold it to - that way it comes under the monthly current budget and not the capital account.
In my experience, very few people use "Xerox" as a verb. I've much more often heard "make some copies on the Xerox machine" (and less often without the "machine") referring generically to a photocopier.
In any case, bad example, as Xerox still holds their trademark.
I can certainly envision someone, having been told to "google it," simply searching using Google or Bing or Dogpile to search.
" preclusion against running 'servers' on residential service."
What's a "server?" A piece of software with a local display and keyboard connecting to the net is called a client if that piece of software is named "web browser" and a server if it is named "X windows." "Server" is an entirely arbitrary distinction.
If the NSA and federal government didn't change after the info was released publicly, why are they acting like an internal complaint might have made a difference?
"a lot of people aren't willing to accept the simple credo of "do good"."
ITYM "Be excellent to each other. And... PARTY ON, DUDES!"
" a landowner may make any legitimate use of their property that they want, even if it interferes with aircraft overflying the land."
So the FAA has no say if I fly drones over my land (or private land where I have the owner's permission).
But, but.... look, shiny!
Too bad it's not removable media.
Try again, with early 1700's technology.
In what way do you consider the choice of a measurement which is easily reproducible virtually anywhere worldwide "arbitrary?"
Yes, things like altitude change the scale a bit. Can you can come up with a better solution (very accessible, reasonably accurate, reasonably reproducible) for transfer of a standard temperature scale worldwide with mid-1700's technology? Choosing the freezing and boiling points of water on that basis for something of scientific, industrial and commercial use seems anything but "arbitrary."
"Why the hell are we talking about the Fahrenheit scale. And, while we're at it, memory of all kinds is always expressed in GiB, so a 512GB card is 1024 times as large as a 512MB card"
I use Rankine, you insensitive clod.
And, you're wrong. The Sandisk 512 GB card being discussed has a capacity of 512,000,000,000 bytes ("1GB=1,000,000,000 bytes" - Sandisk). Just like disk drives and SSDs are measured.
The discussion is about capacity, and disk drives, SSDs, and yes even these memory cards ("1 GB = 1,000,000,000 bytes" - Sandisk) use the metric prefixes correctly. How you can claim that's a "base 10 number of base 2 blocks" is a mystery. Yes, they may address base 2 sized blocks (e.g. 512 or 4096), but the total capacity is specified in base 10. (and the block address is expressed on the interface with a base 2 number, not BCD.) But that's not the capacity, which is what's being discussed.
It seems like disk manufacturers don't reveal specifics like they used to, but it wasn't uncommon to find organizations like 17 sectors/track and other non-base 2/10 layouts.