As for taking back IPv4 addresses, that has to be the most ludicrous thing I have heard. There is a huge amount of IPv4 only content out there which you need IPv4 addresses to reach. Now you can make the consumer IPv6 only by use NAT64 + DNS64 to reach this content but you still need IPv4 addresses on the public side of the NAT64. Additionally NAT64 breaks functionality you get with having direct, unshared, IPv4 connectivity.
About 3.5% of Google's traffic is IPv6. This is more than double what it was last year at this time. If the grow continues on this curve we will be at 10% within a year and a half. This sort of traffic is more than enough for sites to enable IPv6.
If you can enable IPv6 at home over 50% of typical home usage is IPv6 (Google and FaceBook). There is no reason for Consumer ISP's to not enable IPv6 as there is enough volume to make it worthwhile.
IPv6 was designed to co-exist with IPv4. You give new hardware a IPv6 address as well as a IPv4 address and after 10 or so years all your machines have both IPv4 and IPv6 addresses and there are basically no IPv4 only machines left. This would have worked if CPE vendors shipped IPv6 capable routers and ISP enabled IPv6 on the customer links.
Today almost all the cell phones and general purpose computers support both IPv4 and IPv6. Linux, Windows, MacOS, iOS, *BSD, Solaris, Android all support IPv6. What hasn't happened is turning on IPv6 on the ISP links and that is mainly but not solely due to lack of IPv6 in home routers.
Back in 1992 we would have run out except for CIDR and NAT which staved off the exhaustion 20+ years.
People acted on the initial warnings. I was actually running IPv4 networks back then. We upgraded routers and hosts to support CIDR.
More efficient use of addresses was put into play so less addresses were wasted CIDR. By sizing the net mask to the number of hosts on the link we wasted less addresses. Address allocations strategies also changed to assume CIDR was available and you couldn't get more addresses unless you could prove you were properly utilising the addresses you had. If you didn't upgrade your machines to support CIDR you were out of luck if you need new addressees.
Stop gap measures were also put into play like NAT.
Work also started on a replacement for IPv4 which was capable of supporting everyone. That replacement is IPv6.
CIDR and NAT are no longer sufficient to keep IPv4 running.
And of those item which are not available, how many are not there due to new hardware (other than memory and cpu) and how many are not there due to marketing?
FaceTime requires a forward facing camera. Hearing aid support requires new hardware. LTE, HSUPA, nor 802.11n new hardware.
iCloud photo integration marketing. Safari marketing Find My Phone marketing Shared Photo Stream marketing Turn-by-turn navigation marketing Offline Reading List marketing Air Drop marketing Multitasking marketing
Apple want you to buy new hardware. I've got no problem with them not adding new features. If you want the new feature well and good go buy them but not having it doesn't make you horribly obsolete. All of the new stuff falls into the "nice to have" category for 99% of people and given a choice between a 3gs and 5 I would say buy a 5.
Just because Windows didn't have a IP stack 20 years ago, didn't mean you couldn't get OS with a IP stack to run on the hardware. If you look and most operating systems lots of it hasn't changed in 20+ years. 20+ year old bugs are still being found.
Except they are treating the packets differently by allowing vastly different percentages of them the be dropped over different peering links. If the packet is destined for their network and they have control over the link then the packet should be treated equally to all other packets destined for their network. They have (partial) control over the bandwidth of the peering link.
The same also applies to packets leaving their network.
Actually I do expect the desktop to last 10 years. I've got desktop machines that are 15 years old that are still functioning fine for the purpose they are put to. They don't need the extra speed. They chew up a bit of power compared to the more modern hardware that could do the same job.
I had a cable modem that lasted around 12 years. It only got replace this year because it started to stop holding sync. Presumably some capacitor failed in it. I upgraded to a new modem which supported a newer version of DOCSIS. That said I didn't need to upgrade as I didn't need the updated functionality.
As for smart phones becoming obsolete. Most smart phones already have more compute power than is needed. It's nice to have the newer radios and with them faster connections. But apart from some apps that are specifically designed for iOS 7, a 3gs will run everything fine. It has the compute power needed. The screen resolution is good enough for 90% of the population. This is ~5 year old product being first released in 2009. It will do what 90% of the population want to do. It's also still receiving security updates. I applied one over the air (WiFi) within the last month.
The household has iPhone 3gs, 4 and 5 so I've had a chance to evaluate all of them. We max out on the available memory when we buy them. They get replaced due to physical damage. The last 3gs got replaced because it had been dropped to many time and connector #4 would no longer stay seated. My daughter got my wife's 4 and she got a new 5. This was a repair or replace due to physical damage. The phone itself was still capable of doing everything my daughter wanted to do on it.
I reject the contention that in 5 years a smart phone purchased today will be horrendously obsolete because 5 year old phones today are not horrendously obsolete.
Total BS. Phones should last 20 years. The old land line ones last 20+ years. The only thing in a modern phone that doesn't have a 20+ year life span is the battery and that is not through not trying.
As for the 2 years that is the time to pay off the phone in instalments, not when it is supposed to be unusable any more. Yes, phone companies would like you to get a new phone every 2 years as that locks you into them for 2 more years.
As for fixing bugs in the OS most of the time a bug that exists in one version of the OS exist in all versions of the OS. Once the initial diagnoses is done back porting is usually a relatively low cost apart from the regression testing. That said there does become times where back porting becomes expensive. This is usually when a new feature is in the same area of code where the old bug is.
Now in sane countries there are consumer laws about replacement parts needing to be available from the manufacture for reasonable lengths of time for any product being sold. The length of time differs depending upon the product and the price etc. For cars 10-20 years is not unreasonable. Spares are needed to be available well after the warrantee expires.
OS and application bug fixes are no more than getting a spare parts and should be available for similar lengths of time. The only reason they aren't is that consumer law hasn't caught up yet.
Only if the cost of maintain the police force exceeds the saving in running hospitals etc. Driverless cars should be a net gain for society. If your government is not structured to achieve that gain you should adjust your system of government.
For somewhere like NSW where the state government runs most of the hospitals that deal with road trauma and also runs the police force that enforces speed limits driverless cars will be a net saving to the state budget.
Even if both parties have iPhones there are times when you want to force SMS messages and iMessage doesn't have a simple interface to do that. You have to go into settings and turn off iMessage if you want to force a SMS.
You just do time limited downloads. 1 week rental $A. 2 weeks rental $A + $B. 4 weeks rental $A + $B +$C.
The cable company here has 48hr on demand. After 48hrs you have to pony up to watch it again but within that 48 hours you can watch it as many times as you like.
And every building built on this planet contains defects. Every bridge built on this planet contains defects. It is impossible to build a building or bridge without there being a defect. Almost all the the defects are insignificant. The difference between Software and Civil Engineering that small defects don't usually bring the whole construction down whereas the smallest defect in a piece of software often have catastrophic effects.
Then you should complain to the router vendor. Dynamic UPDATE and TSIG were standardised over a decade ago to allow everyone to use the same protocol. You should just be able to type in the update server's name if it isn't one of the zone's nameservers. If it is then it shouldn't be needed.
nsupdate server <servername> key keyname <secret> update delete hostname A update add hostname 300 A 1.2.3.4 send
Actually the Internet roads haven't changed much. Its been twisted pair copper (telegraph -> 110bd modems -> DSL (last 80 years)), some coax (various stuff including various flavours of DOCIS (last 50 years), now fibre (which has seen the frequency increase as the optics at the ends improve (last 30 years)) and a bit of wireless. What has changed is the modulation of the data (cars) transmitted over it.
Now you can lease fibres or you can lease layer 2 access to those fibres (or do a mix). In either case you get more speed by changing the technology at the ends (cars) not by changing the fibre (roads). There is no clear winner over which type of lease to do.
Standardising the technology at each end is what has made the internet grow and it has made the technology affordable for just about everyone.
Actually different nameservers are tolerant of different errors. Things like turning on IPv6 (which Comcast does) make various configuration mistakes visible which would otherwise usually remain hidden.
And you evidence for this is what? That name runs sanity checks on the function arguments then dies if it finds that the contract is not met? If system integrators did what Apple did and run named from launchd or the equivalent, which restarts named on unexpected terminations, there would be almost no advisories as the availability impact would go from complete (7.8) to partial (5.0) with even the easiest denial of service flaws and drops down still further to 4.3 or 2.6 with the more complicated ones. However the assumption is that there is nothing restarting named so a advisory with a 7.8, 7.1 or 5.4 score is issued.
Please go read the descriptions in the advisories at BIND-9-Security-Vulnerability-Matrix then ask yourself do you want software written by people who know they are human so mistakes will be made and therefor check for them or software that assumes that everything will always be correct and continues of regardless of the garbage arguments it has been given.
Requiring 1:1 traffic flows is a stupid way to organise peering agreements. It doesn't matter if it is 1000:1 as long as both sides are getting benefit from the peering. If you have content you want to deliver and eyeballs that want content you peer.
As for transit providers and eyeball networks. The transit providers are being paid to deliver content and the eyeball networks are being paid to deliver content. Failing to upgrade interconnects is taking monies under false pretences and that has another name "fraud".
The last mile is essentially a one off cost if you are using your own infrastructure. There is the occasional repairs due to failed equipment and backhoe / tree branch fade. These costs are essentially bandwidth independent. As technology changes you have the occasional replacement of head end equipment to support new line protocols over the existing cabling. The costs of upgrading head end equipment can often be absorbed into the maintenance budget by replacing failed components with components that have newer functionality.
The trunks are a recurrent which continually need to be upgraded to meet the increasing aggregate usage of all the customers either by replacing the optics or lighting more fibres which may involve trenching. For trunks there are also transit costs.
Data centres are a bit like both. The bandwidth is essentially free but there are huge costs when you exceed the physical capacity.
If you thing SPF as made a difference to the amount of spam being sent I have a bridge to sell you.
SPF was never about preventing spam. It has only ever been about preventing your email address being used as the from address in spam. It reduces the amount of blowback to your account and nothing else. The only reason SPF appears to be a effective anti-spam tool is that there is that the number of sites filtering using SPF hasn't risen to the level where the spammers need to stop using SPF protected address as the from addresses.
Another thing SPF does is cause spammers to use hijacked credentials to send spam through legitimate sources.
This shouldn't be seen as a reason to stop using SPF records. But if you think SPF will stop spam you are deluding yourself.
It's very much dependent on the current path. Hand to foot and you have a good chance of survival. Hand to hand and your survival rate drops. Additionally whether you pick up the live object or brush the live object affects your survival rate.
Having brushed a live 240v connector with the back of my hand, and survived, I never want to do that again.
Slashdot Mirror
255.255.255.255.255.255.255.255.255.255.255.255.255.255.255.255 vs ffff:ffff:ffff:ffff:ffff:ffff:ffff:ffff
Which would you rather type, read etc.
Comcast has over 25% of their network IPv6 enabled as of November last year. This is much more that "3 cities".
As for taking back IPv4 addresses, that has to be the most ludicrous thing I have heard. There is a huge amount of IPv4 only content out there which you need IPv4 addresses to reach. Now you can make the consumer IPv6 only by use NAT64 + DNS64 to reach this content but you still need IPv4 addresses on the public side of the NAT64. Additionally NAT64 breaks functionality you get with having direct, unshared, IPv4 connectivity.
About 3.5% of Google's traffic is IPv6. This is more than double what it was last year at this time. If the grow continues on this curve we will be at 10% within a year and a half. This sort of traffic is more than enough for sites to enable IPv6.
If you can enable IPv6 at home over 50% of typical home usage is IPv6 (Google and FaceBook). There is no reason for Consumer ISP's to not enable IPv6 as there is enough volume to make it worthwhile.
IPv6 was designed to co-exist with IPv4. You give new hardware a IPv6 address as well as a IPv4 address and after 10 or so years all your machines have both IPv4 and IPv6 addresses and there are basically no IPv4 only machines left. This would have worked if CPE vendors shipped IPv6 capable routers and ISP
enabled IPv6 on the customer links.
Today almost all the cell phones and general purpose computers support both IPv4 and IPv6. Linux, Windows, MacOS, iOS, *BSD, Solaris, Android all support IPv6. What hasn't happened is turning on IPv6 on the ISP links and that is mainly but not solely due to lack of IPv6 in home routers.
Back in 1992 we would have run out except for CIDR and NAT which staved off the exhaustion 20+ years.
People acted on the initial warnings. I was actually running IPv4 networks back then. We upgraded routers and hosts to support CIDR.
More efficient use of addresses was put into play so less addresses were wasted CIDR. By sizing the net mask to the number of hosts on the link we wasted less addresses. Address allocations strategies also changed to assume CIDR was available and you couldn't get more addresses unless you could prove you were properly utilising the addresses you had. If you didn't upgrade your machines to support CIDR you were out of luck if you need new addressees.
Stop gap measures were also put into play like NAT.
Work also started on a replacement for IPv4 which was capable of supporting everyone. That replacement is IPv6.
CIDR and NAT are no longer sufficient to keep IPv4 running.
And of those item which are not available, how many are not there due to new hardware (other than memory and cpu) and how many are not there due to marketing?
FaceTime requires a forward facing camera.
Hearing aid support requires new hardware.
LTE, HSUPA, nor 802.11n new hardware.
iCloud photo integration marketing.
Safari marketing
Find My Phone marketing
Shared Photo Stream marketing
Turn-by-turn navigation marketing
Offline Reading List marketing
Air Drop marketing
Multitasking marketing
Apple want you to buy new hardware. I've got no problem with them not adding new features. If you want the new feature well and good go buy them but not having it doesn't make you horribly obsolete. All of the new stuff falls into the "nice to have" category for 99% of people and given a choice between a 3gs and 5 I would say buy a 5.
Just because Windows didn't have a IP stack 20 years ago, didn't mean you couldn't get OS with a IP stack to run on the hardware. If you look and most operating systems lots of it hasn't changed in 20+ years. 20+ year old bugs are still being found.
Except they are treating the packets differently by allowing vastly different percentages of them the be dropped over different peering links. If the packet is destined for their network and they have control over the link then the packet should be treated equally to all other packets destined for their network. They have (partial) control over the bandwidth of the peering link.
The same also applies to packets leaving their network.
Actually I do expect the desktop to last 10 years. I've got desktop machines that are 15 years old that are still functioning fine for the purpose they are put to. They don't need the extra speed. They chew up a bit of power compared to the more modern hardware that could do the same job.
I had a cable modem that lasted around 12 years. It only got replace this year because it started to stop holding sync. Presumably some capacitor failed in it. I upgraded to a new modem which supported a newer version of DOCSIS. That said I didn't need to upgrade as I didn't need the updated functionality.
As for smart phones becoming obsolete. Most smart phones already have more compute power than is needed. It's nice to have the newer radios and with them faster connections. But apart from some apps that are specifically designed for iOS 7, a 3gs will run everything fine. It has the compute power needed. The screen resolution is good enough for 90% of the population. This is ~5 year old product being first released in 2009. It will do what 90% of the population want to do. It's also still receiving security updates. I applied one over the air (WiFi) within the last month.
The household has iPhone 3gs, 4 and 5 so I've had a chance to evaluate all of them. We max out on the available memory when we buy them. They get replaced due to physical damage. The last 3gs got replaced because it had been dropped to many time and connector #4 would no longer stay seated. My daughter got my wife's 4 and she got a new 5. This was a repair or replace due to physical damage. The phone itself was still capable of doing everything my daughter wanted to do on it.
I reject the contention that in 5 years a smart phone purchased today will be horrendously obsolete because 5 year old phones today are not horrendously obsolete.
Total BS. Phones should last 20 years. The old land line ones last 20+ years. The only thing in a modern phone that doesn't have a 20+ year life span is the battery and that is not through not trying.
As for the 2 years that is the time to pay off the phone in instalments, not when it is supposed to be unusable any more. Yes, phone companies would like you to get a new phone every 2 years as that locks you into them for 2 more years.
As for fixing bugs in the OS most of the time a bug that exists in one version of the OS exist in all versions of the OS. Once the initial diagnoses is done back porting is usually a relatively low cost apart from the regression testing. That said there does become times where back porting becomes expensive. This is usually when a new feature is in the same area of code where the old bug is.
Now in sane countries there are consumer laws about replacement parts needing to be available from the manufacture for reasonable lengths of time for any product being sold. The length of time differs depending upon the product and the price etc. For cars 10-20 years is not unreasonable. Spares are needed to be available well after the warrantee expires.
OS and application bug fixes are no more than getting a spare parts and should be available for similar lengths of time. The only reason they aren't is that consumer law hasn't caught up yet.
Actually they are peering even in the co-lo. The peering connection(s) may go to a single machine(s) but they are peering.
Because it provides a incentive to corrupt practices.
Only if the cost of maintain the police force exceeds the saving in running hospitals etc. Driverless cars should be a net gain for society. If your government is not structured to achieve that gain you should adjust your system of government.
For somewhere like NSW where the state government runs most of the hospitals that deal with road trauma and also runs the police force that enforces speed limits driverless cars will be a net saving to the state budget.
Even if both parties have iPhones there are times when you want to force SMS messages and iMessage doesn't have a simple interface to do that. You have to go into settings and turn off iMessage if you want to force a SMS.
DANE
You just do time limited downloads. 1 week rental $A. 2 weeks rental $A + $B. 4 weeks rental $A + $B +$C.
The cable company here has 48hr on demand. After 48hrs you have to pony up to watch it again but within that
48 hours you can watch it as many times as you like.
And every building built on this planet contains defects. Every bridge built on this planet contains defects. It is impossible to build a building or bridge without there being a defect. Almost all the the defects are insignificant. The difference between Software and Civil Engineering that small defects don't usually bring the whole construction down whereas the smallest defect in a piece of software often have catastrophic effects.
Then you should complain to the router vendor. Dynamic UPDATE and TSIG were
standardised over a decade ago to allow everyone to use the same protocol.
You should just be able to type in the update server's name if it isn't one of
the zone's nameservers. If it is then it shouldn't be needed.
nsupdate
server <servername>
key keyname <secret>
update delete hostname A
update add hostname 300 A 1.2.3.4
send
Actually the Internet roads haven't changed much. Its been twisted pair copper (telegraph -> 110bd modems -> DSL
(last 80 years)), some coax (various stuff including various flavours of DOCIS (last 50 years), now fibre (which has
seen the frequency increase as the optics at the ends improve (last 30 years)) and a bit of wireless. What has
changed is the modulation of the data (cars) transmitted over it.
Now you can lease fibres or you can lease layer 2 access to those fibres (or do a mix). In either case you get
more speed by changing the technology at the ends (cars) not by changing the fibre (roads). There is no clear
winner over which type of lease to do.
Standardising the technology at each end is what has made the internet grow and it has made the technology
affordable for just about everyone.
Actually different nameservers are tolerant of different errors. Things like turning on IPv6 (which Comcast does) make various configuration mistakes visible which would otherwise usually remain hidden.
And you evidence for this is what? That name runs sanity checks on the function arguments then dies if it finds that the contract is not met? If system integrators did what Apple did and run named from launchd or the equivalent, which restarts named on unexpected terminations, there would be almost no advisories as the availability impact would go from complete (7.8) to partial (5.0) with even the easiest denial of service flaws and drops down still further to 4.3 or 2.6 with the more complicated ones. However the assumption is that there is nothing restarting named so a advisory with a 7.8, 7.1 or 5.4 score is issued.
Please go read the descriptions in the advisories at BIND-9-Security-Vulnerability-Matrix then ask yourself do you want software written by people who know they are human so mistakes will be made and therefor check for them or software that assumes that everything will always be correct and continues of regardless of the garbage arguments it has been given.
Requiring 1:1 traffic flows is a stupid way to organise peering agreements. It doesn't matter if it is 1000:1 as long as both sides are getting benefit from the peering. If you have content you want to deliver and eyeballs that want content you peer.
As for transit providers and eyeball networks. The transit providers are being paid to deliver content and the eyeball networks are being paid to deliver content. Failing to upgrade interconnects is taking monies under false pretences and that has another name "fraud".
The last mile is essentially a one off cost if you are using your own infrastructure. There is the occasional repairs due to failed equipment and backhoe / tree branch fade. These costs are essentially bandwidth independent. As technology changes you have the occasional replacement of head end equipment to support new line protocols over the existing cabling. The costs of upgrading head end equipment can often be absorbed into the maintenance budget by replacing failed components with components that have newer functionality.
The trunks are a recurrent which continually need to be upgraded to meet the increasing aggregate usage of all the customers either by replacing the optics or lighting more fibres which may involve trenching. For trunks there are also transit costs.
Data centres are a bit like both. The bandwidth is essentially free but there are huge costs when you exceed the physical capacity.
If you thing SPF as made a difference to the amount of spam being sent I have a bridge to sell you.
SPF was never about preventing spam. It has only ever been about preventing your email address being used as the from address in spam. It reduces the amount of blowback to your account and nothing else. The only reason SPF appears to be a effective anti-spam tool is that there is that the number of sites filtering using SPF hasn't risen to the level where the spammers need to stop using SPF protected address as the from addresses.
Another thing SPF does is cause spammers to use hijacked credentials to send spam through legitimate sources.
This shouldn't be seen as a reason to stop using SPF records. But if you think SPF will stop spam you are deluding yourself.
Sheep.
Fish.
Bacteria.
Not all plurals end in 's'.
It's very much dependent on the current path. Hand to foot and you have a good chance of survival. Hand to hand and your survival rate drops. Additionally whether you pick up the live object or brush the live object affects your survival rate.
Having brushed a live 240v connector with the back of my hand, and survived, I never want to do that again.