Ditching it in the sea and recovering it causes too much damage to make it viable to refit. This was intended for the boosters on the space shuttle, but it ended up being cheaper to make new ones than fix the old ones.
Of course they could bring them down over land, but I think the unpredictability of exactly where they would land could be marginally terrifying.
That was using multiple wavelengths on multiple fibres. This appears to be one wavelength on one fibre. Different kettle of fish.
"Researchers from the NEC Labs in Princeton, NJ, USA, and from Corningâ(TM)s Sullivan Park Research Center in Corning, NY, successfully demonstrated ultra-high speed transmission with a capacity of 1.05 petabit/s (1015 bits per second) over novel multi-core fiber that contains 12 single-mode and two few-mode cores by employing the advanced space division multiplexing scheme and optical multiple-input multiple-output signal processing technique."
"The superchannel is an advanced dense wave division multiplexing (DWDM) technique, created by combining multiple coherent optical signals into one channel"
Not quite. Those optics use DP-QPSK, which uses mathematical magic to cram 4 bits worth of information into one symbol. This means the optics only need to operate at 25Gbps to supply a 100Gbps line rate.
DP-QPSK is a whole load of magic I don't understand.
If DP-QPSK can be used with this technology, it seems to imply 200Gbps optics are not too far away.
So now, you have just the global prefix space. Of this, the first 2 bytes are assigned by the IANA to the RIR - the 2001, the 2400.... It's not a part of what your RIR gets to give you.
I've been trying to work out what you mean by all this. IANA doesn't allocate to RIRs on/16 (2 bytes) boundaries. It allocates based on whatever is appropriate. The latest allocation was a/12 to ARIN. I don't see that it has ever allocated a/16 either. (a/16 was reserved for 6to4) http://www.iana.org/assignment...
Now, depending on the geographic reach of the ISP, they may need thousands of offices nationwide, and in each office, service several thousands of people. Let us assume that we have 16 million routers serviced that way - that is 2^14. So your 32 bits are now down to 18. So it now comes down to how many people are serviced by a single central office router. Lets assume it's 128, which is 2^7, and you are down to 11.
This really doesn't make sense to me. Firstly 2^14 is around 16k, not 16 million... What are these routers that you are servicing? CPEs? In this scenario you require 16 million/48s which would fit into a/24 quite nicely.
So you are already cutting into the subnet address space of the IPv6 address, since you have only 11 bits to give a customer for subnetting. Giving everyone a/48, as you mentioned, would give each customer 16 bits of subnetting address, so you now have a deficit of 5. Which is why RIRs like APNIC and RIPE assign/56s instead. Each customer that way gets 8 bits of subnetting, instead of 16. Which may or may not be adequate.
I genuinely don't understand what you are talking about. Maybe a diagram would help? You mention an inherent structure in v6 addresses. Maybe that is where you are getting confused, because other than the 64:64 split, there is none as such...
So now, you have just the global prefix space. Of this, the first 2 bytes are assigned by the IANA to the RIR - the 2001, the 2400.... It's not a part of what your RIR gets to give you. The best your RIR can give you is a/32, and you had better be a huuuuugge organization - maybe global - w/ millions of SUBNETS to justify that. So now the ISP has 48 bits, down from the 64 - 2001:db8:b10c:abcd::/64.
...
So you are already cutting into the subnet address space of the IPv6 address, since you have only 11 bits to give a customer for subnetting. Giving everyone a/48, as you mentioned, would give each customer 16 bits of subnetting address, so you now have a deficit of 5. Which is why RIRs like APNIC and RIPE assign/56s instead. Each customer that way gets 8 bits of subnetting, instead of 16. Which may or may not be adequate.
A lot of this is incorrect. RIPE by default allocate a/29 to ISPs. Getting something larger than that is super easy, as long as you have the documentation to back it up.
Also RIRs don't assign/56s to customers. The RIR allocates a prefix to a LIR (Your ISP) and the LIR allocates the addresses down to the customer.
And finally the RIPE policy (and likely others. I live and work in the RIPE region so my knowledge is more relevant to that region) recommends a/48 for end user allocation. (https://www.ripe.net/publications/docs/ripe-655. If you read from section 5.3 onwards you will see that it is up to the ISP to decide what to allocate to a customer. Anything shorter than a/48 requires documentation, but a/48 is just fine if you want. The wording has actually changed on that page recently. It used to be more specific about recommendation of prefix sizes.
There is nothing inherently special about any of the top 64 bits. While they are divvied out to RIRs by reserving so many bits from the top, it doesn't break the maths that there is a fuck tonne of/48's available. Even with outlandish allocation policies it is unlikely to ever run out. At least not in the useful life of the protocol. We are bound to hit some limitations with IPv6, but it was designed to not be the size of the address space. Why stifle potential innovation?
And splitting the address space any other way was never an option. It was always going to be 64 bits at the top, and there was only ever a question about the size of the bottom part of the address.
Not giving everyone a/48 is a daft argument. From someone who is a lot smarter than me source
"Let’s assume that ISPs come in essentially 3 flavors. MEGA (The Verizons, AT&Ts, Comcasts, etc. of the world) having more than 5 million customers, LARGE (having between 100,000and 5 million customers) and SMALL (having fewer than 100,000 customers).
Let’s assume the worst possible splits and add 1 nibble to the minimum needed for each ISP and another nibble for overhead.
Further, let’s assume that 7 billion people on earth all live in individual households and that each of them runs their own small business bringing the total customer base worldwide to 14 billion.
If everyone subscribes to a MEGA and each MEGA serves 5 million customers, we need 2,800 MEGA ISPs. Each of those will need 5,000,000/48s which would require a/24. Let’s give each of those an additional 8 bits for overhead and bad splits and say each of them gets a/16. That’s 2,800 out of 65,536/16s and we’ve served every customer on the planet with a lot of extra overhead, using approximately 4% of the address space.
Now, let’s make another copy of earth and serve everyone on a LARGE ISP with only 100,000 customers each. This requires 140,000 LARGE ISPs each of whom will need a/28 (100,000/48s doesn’t fit in a/32, so we bump them up to/28). Adding in bad splits and overhead at a nibble each, we give each of them a/20. 140,000/20s out of 1,048,576 total of which we used 44,800 for the MEGA ISPS leaves us with 863,776/20s still available. We’ve now managed to burn approximately 18% of the total address space and we’ve served the entire world twice.
Finally, let us serve every customer in the world using a small ISP. Let’s assume that each small ISP only serves about 5,000 customers. For 5,000 customers, we would need a/32. Backing that off two nibbles for bad splits and overhead, we give each one a/24.
This will require 2,800,000/24s. (I realize lots of ISPs server fewer than 5,000 customers, but those ISPs also don’t serve a total of 14 billion end sites, so I think in terms of averages, this is not an unreasonable place to throw the dart).
There are 16,777,216/24s in total, but we’ve already used 2,956,800 for the MEGA and LARGE ISPs, bringing our total utilization to 5,756,800/24s.
We have now built three complete copies of the internet with some really huge assumptions about number of households and businesses added in and we still have only used roughly 34% of the total address space, including nibble boundary round-ups and everything else."
Is that the metric that keeps IPv6 adaption capped?
I asked the owner of an ISP how he was going to deal with IPv6. His answer was, "Buy a lot of expensive hardware." That is the metric that keeps IPv6 adoption capped: people don't want to pay for new hardware.
As someone who works for ISPs for a living, that is nonsense. Equipment generally has a lifetime that it is useful for. We typically buy kit with 5 years in mind, but may stretch it further if there is still life in it. Equipment that is 10 years old is probably worthless (This likely is the same for most other areas of IT)
Any equipment you buy today will support IPv6, with all the latest standards. Equipment generally gets firmware upgrades for the duration of its life that adds new features as they come along.
All Cisco and Juniper kit (2 big vendors in the ISP space) have had full feature sets for v6 in the service provider routed world for quite some time now. So long that some of their kit has gone end of life that have v6 support. There may be some enterprise grade products where this doesn't hold true, but it shouldn't be far off.
If your friend claims that the way he is going to deal with v6 is to buy more kit, he is either running outdated equipment, stupid, or lying.
The CPE is the only major space where there is issues. This is getting better now, and the same 5 year rule generally applies here to ageing equipment. You have the luxury of a phased replacement plan in this space too, which makes things a bit simpler.
Exactly. People who run NAT as a firewall and think there is no security in IPv6 because there is no NAT is dumb. The simple basic firewall that should come on all CPE is default deny inbound, and a reflective permit outbound. This will give the actual security that people think they have with NAT.
Back before the exhaustion policies kicked in, ARIN were burning through a/8 every couple of months.
This is why taking back the legacy address allocations will not really be worth the time or effort. There is more demand than availability. If there was free reign allocation over it all, it would be gone before the year is out.
A local internet registry at smallest only gets a/32.
This really depends on your region. My knowledge if from RIPE. The default allocation there is now a/29. But that is default. If you can justify more, you can have more.
I'm a fan of a/60 for homes. I guess you are right there is enough room to make a/48 work but that seems like needlessly throwing away a lot of bits.
Partially from a previous comment I've made:
Imagine that everyone on the planet is connected, and they each have 32 different ISPs (phones, home, work...) This is a gross overestimation.
This easily fits within a single/10. That is 1/1024 of the total address space.
IPv4 currently has been around since around 1980 (Can't be bothered to get real dates). This means that it has so far had a life of around 40 years.
IPv6 is not going to last forever. It is very likely we will hit some limitation of the protocol, but its not likely to be with the address space. Lets use the IPv4 life length as a ball park figure for how long its going to last us.
Lets imagine in that time the population doubles, and the number of ISPs that everyone has doubles in that time.
This comfortably fits within a/8 or 1/256 of the available address space.
Now we can do sparse addressing and leave big holes in the allocations "just in case" but we are still going to have a hell of a lot of address space left at this point.
Since we don't expect this protocol to last forever, why potentially stifle innovation by limiting addresses, when even using really outlandish figures for what may happen still leaves us with huge swaths of address space unused?
The current best practice allocation policies only affect the first/3, or 1/8 of the total available space. If we manage to burn through that quicker than expected, policies can be adjusted for the next/3.
I will begrudgingly accept for an ISP to hand me a/60 or a/56, but personally, I will be giving all my customers a/48. The space is so massive it seems rude not to.
If you run a business and need your address space to never change, like when changing ISP, then PI (provider independent) address space is the way to go. The address space will always be yours, and you can take it with you.
For residential users who like to know where their devices are... that one is a bit tougher. Really DNS is your friend. How often do you change ISPs anyway?
"So a/48 should be used when there is any doubt whether a/56 is sufficient in the long run. ISPs get much leeway in determining the prefix size they give to their customers up to/48–even in the case of home users"
I would say there is always a doubt that a/56 may be insufficient. A/56 only allows for 256 networks.
Slashdot Mirror
I meant bringing them back with parachutes. A burn back is much more controlled.
The damage is caused by the salt in the water. Missed that out!
Ditching it in the sea and recovering it causes too much damage to make it viable to refit. This was intended for the boosters on the space shuttle, but it ended up being cheaper to make new ones than fix the old ones.
Of course they could bring them down over land, but I think the unpredictability of exactly where they would land could be marginally terrifying.
Yes, QPSK does send 2 symbols, but DP-QPSK doubles that to 4 symbols.
Multiple cores, multiple fibres. Effectively the same thing. I believe this test was over 1 core. That's the big difference.
That was using multiple wavelengths on multiple fibres. This appears to be one wavelength on one fibre. Different kettle of fish.
"Researchers from the NEC Labs in Princeton, NJ, USA, and from Corningâ(TM)s Sullivan Park Research Center in Corning, NY, successfully demonstrated ultra-high speed transmission with a capacity of 1.05 petabit/s (1015 bits per second) over novel multi-core fiber that contains 12 single-mode and two few-mode cores by employing the advanced space division multiplexing scheme and optical multiple-input multiple-output signal processing technique."
Nope, that is 100Gbase-(L/S/E)R4. ZR uses DP-QSPK.
"The superchannel is an advanced dense wave division multiplexing (DWDM) technique, created by combining multiple coherent optical signals into one channel"
Not quite. Those optics use DP-QPSK, which uses mathematical magic to cram 4 bits worth of information into one symbol. This means the optics only need to operate at 25Gbps to supply a 100Gbps line rate.
DP-QPSK is a whole load of magic I don't understand.
If DP-QPSK can be used with this technology, it seems to imply 200Gbps optics are not too far away.
Google will fix this by updating the ASOP with their fix in the latest version. Possibly even a few previous versions too.
The problem is the handset makers and carriers won't push the updates down to the handsets they support.
So now, you have just the global prefix space. Of this, the first 2 bytes are assigned by the IANA to the RIR - the 2001, the 2400.... It's not a part of what your RIR gets to give you.
I've been trying to work out what you mean by all this. /16 (2 bytes) boundaries. It allocates based on whatever is appropriate. The latest allocation was a /12 to ARIN. I don't see that it has ever allocated a /16 either. (a /16 was reserved for 6to4) http://www.iana.org/assignment...
IANA doesn't allocate to RIRs on
Now, depending on the geographic reach of the ISP, they may need thousands of offices nationwide, and in each office, service several thousands of people. Let us assume that we have 16 million routers serviced that way - that is 2^14. So your 32 bits are now down to 18. So it now comes down to how many people are serviced by a single central office router. Lets assume it's 128, which is 2^7, and you are down to 11.
This really doesn't make sense to me. Firstly 2^14 is around 16k, not 16 million... What are these routers that you are servicing? CPEs? In this scenario you require 16 million /48s which would fit into a /24 quite nicely.
So you are already cutting into the subnet address space of the IPv6 address, since you have only 11 bits to give a customer for subnetting. Giving everyone a /48, as you mentioned, would give each customer 16 bits of subnetting address, so you now have a deficit of 5. Which is why RIRs like APNIC and RIPE assign /56s instead. Each customer that way gets 8 bits of subnetting, instead of 16. Which may or may not be adequate.
I genuinely don't understand what you are talking about. Maybe a diagram would help? You mention an inherent structure in v6 addresses. Maybe that is where you are getting confused, because other than the 64:64 split, there is none as such...
So now, you have just the global prefix space. Of this, the first 2 bytes are assigned by the IANA to the RIR - the 2001, the 2400.... It's not a part of what your RIR gets to give you. The best your RIR can give you is a /32, and you had better be a huuuuugge organization - maybe global - w/ millions of SUBNETS to justify that. So now the ISP has 48 bits, down from the 64 - 2001:db8:b10c:abcd::/64.
So you are already cutting into the subnet address space of the IPv6 address, since you have only 11 bits to give a customer for subnetting. Giving everyone a /48, as you mentioned, would give each customer 16 bits of subnetting address, so you now have a deficit of 5. Which is why RIRs like APNIC and RIPE assign /56s instead. Each customer that way gets 8 bits of subnetting, instead of 16. Which may or may not be adequate.
A lot of this is incorrect. RIPE by default allocate a /29 to ISPs. Getting something larger than that is super easy, as long as you have the documentation to back it up.
Also RIRs don't assign /56s to customers. The RIR allocates a prefix to a LIR (Your ISP) and the LIR allocates the addresses down to the customer.
And finally the RIPE policy (and likely others. I live and work in the RIPE region so my knowledge is more relevant to that region) recommends a /48 for end user allocation. (https://www.ripe.net/publications/docs/ripe-655. If you read from section 5.3 onwards you will see that it is up to the ISP to decide what to allocate to a customer. Anything shorter than a /48 requires documentation, but a /48 is just fine if you want. The wording has actually changed on that page recently. It used to be more specific about recommendation of prefix sizes.
There is nothing inherently special about any of the top 64 bits. While they are divvied out to RIRs by reserving so many bits from the top, it doesn't break the maths that there is a fuck tonne of /48's available. Even with outlandish allocation policies it is unlikely to ever run out. At least not in the useful life of the protocol. We are bound to hit some limitations with IPv6, but it was designed to not be the size of the address space. Why stifle potential innovation?
And splitting the address space any other way was never an option. It was always going to be 64 bits at the top, and there was only ever a question about the size of the bottom part of the address.
Maybe that was the case on some super old Cisco kit. Anything bought in the past 5 years at least has forwarding in hardware for all packets.
Not giving everyone a /48 is a daft argument. From someone who is a lot smarter than me source
"Let’s assume that ISPs come in essentially 3 flavors. MEGA (The Verizons, AT&Ts, Comcasts, etc. of the world) having more than 5 million customers, LARGE (having between 100,000and 5 million customers) and SMALL (having fewer than 100,000 customers).
Let’s assume the worst possible splits and add 1 nibble to the minimum needed for each ISP and another nibble for overhead.
Further, let’s assume that 7 billion people on earth all live in individual households and that each of them runs their own small business bringing the total customer base worldwide to 14 billion.
If everyone subscribes to a MEGA and each MEGA serves 5 million customers, we need 2,800 MEGA ISPs. Each of those will need 5,000,000 /48s which would require a /24. Let’s give each of those an additional 8 bits for overhead and bad splits and say each of them gets a /16. That’s 2,800 out of /16s and we’ve served every customer on the planet with a lot of extra overhead, using approximately 4% of the address space.
65,536
Now, let’s make another copy of earth and serve everyone on a LARGE ISP with only 100,000 customers each. This requires 140,000 LARGE ISPs each of whom will need a /28 (100,000 /48s doesn’t fit in a /32, so we bump them up to /28). Adding in bad splits and overhead at a nibble each, we give each of them a /20. 140,000 /20s out of 1,048,576 total of which we used 44,800 for the MEGA ISPS leaves us with 863,776 /20s still available. We’ve now managed to burn approximately 18% of the total address space and we’ve served the entire world twice.
Finally, let us serve every customer in the world using a small ISP. Let’s assume that each small ISP only serves about 5,000 customers. For 5,000 customers, we would need a /32. Backing that off two nibbles for bad splits and overhead, we give each one a /24.
This will require 2,800,000 /24s. (I realize lots of ISPs server fewer than 5,000 customers, but those ISPs also don’t serve a total of 14 billion end sites,
so I think in terms of averages, this is not an unreasonable place to throw the dart).
There are 16,777,216 /24s in total, but we’ve already used 2,956,800 for the MEGA and LARGE ISPs, bringing our total utilization to 5,756,800 /24s.
We have now built three complete copies of the internet with some really huge assumptions about number of households and businesses added in and we still have only used roughly 34% of the total address space, including nibble boundary round-ups and everything else."
Is that the metric that keeps IPv6 adaption capped?
I asked the owner of an ISP how he was going to deal with IPv6. His answer was, "Buy a lot of expensive hardware." That is the metric that keeps IPv6 adoption capped: people don't want to pay for new hardware.
As someone who works for ISPs for a living, that is nonsense. Equipment generally has a lifetime that it is useful for. We typically buy kit with 5 years in mind, but may stretch it further if there is still life in it. Equipment that is 10 years old is probably worthless (This likely is the same for most other areas of IT)
Any equipment you buy today will support IPv6, with all the latest standards. Equipment generally gets firmware upgrades for the duration of its life that adds new features as they come along.
All Cisco and Juniper kit (2 big vendors in the ISP space) have had full feature sets for v6 in the service provider routed world for quite some time now. So long that some of their kit has gone end of life that have v6 support. There may be some enterprise grade products where this doesn't hold true, but it shouldn't be far off.
If your friend claims that the way he is going to deal with v6 is to buy more kit, he is either running outdated equipment, stupid, or lying.
The CPE is the only major space where there is issues. This is getting better now, and the same 5 year rule generally applies here to ageing equipment. You have the luxury of a phased replacement plan in this space too, which makes things a bit simpler.
Exactly. People who run NAT as a firewall and think there is no security in IPv6 because there is no NAT is dumb. The simple basic firewall that should come on all CPE is default deny inbound, and a reflective permit outbound. This will give the actual security that people think they have with NAT.
Absolutely. The following sentence on their about us page says it all:
"CHM Construction By Simonds Homes, Australia's Legacy of 65 years in Constructions and Public listing Builder Group."
They are claiming to be the same company.
Why do you need multiple IPs on a machine with 1 nic? Besides the last 64 is going to be the MAC address.
The real question is "Why wouldn't you?"
Giving machines the ability to have multiple IP addresses opens up possibilities for people.
Only 5? You're not thinking big enough!
As many IPs as you want per computer. Go nuts!
Back before the exhaustion policies kicked in, ARIN were burning through a /8 every couple of months.
This is why taking back the legacy address allocations will not really be worth the time or effort. There is more demand than availability. If there was free reign allocation over it all, it would be gone before the year is out.
Move to IPv6 already.
Oh, and 11/8 recently became routable.
AFAIk the 2001:678::/29 provider independent space are just supposed to be a bunch of /48 relays.
What now? PI space is normal routable space that you take between providers. Nothing to do with relays...
https://en.wikipedia.org/wiki/...
A local internet registry at smallest only gets a /32.
This really depends on your region. My knowledge if from RIPE. The default allocation there is now a /29. But that is default. If you can justify more, you can have more.
I'm a fan of a /60 for homes. I guess you are right there is enough room to make a /48 work but that seems like needlessly throwing away a lot of bits.
Partially from a previous comment I've made:
Imagine that everyone on the planet is connected, and they each have 32 different ISPs (phones, home, work...) This is a gross overestimation.
7 billion people * 32 = 224 000 000 000 /48's required.
This easily fits within a single /10. That is 1/1024 of the total address space.
IPv4 currently has been around since around 1980 (Can't be bothered to get real dates). This means that it has so far had a life of around 40 years.
IPv6 is not going to last forever. It is very likely we will hit some limitation of the protocol, but its not likely to be with the address space. Lets use the IPv4 life length as a ball park figure for how long its going to last us.
Lets imagine in that time the population doubles, and the number of ISPs that everyone has doubles in that time.
We now have 14 billion people and 64 ISPs
14 billion people * 64 = 896 000 000 000 /48s required.
This comfortably fits within a /8 or 1/256 of the available address space.
Now we can do sparse addressing and leave big holes in the allocations "just in case" but we are still going to have a hell of a lot of address space left at this point.
Since we don't expect this protocol to last forever, why potentially stifle innovation by limiting addresses, when even using really outlandish figures for what may happen still leaves us with huge swaths of address space unused?
The current best practice allocation policies only affect the first /3, or 1/8 of the total available space. If we manage to burn through that quicker than expected, policies can be adjusted for the next /3.
I will begrudgingly accept for an ISP to hand me a /60 or a /56, but personally, I will be giving all my customers a /48. The space is so massive it seems rude not to.
If you run a business and need your address space to never change, like when changing ISP, then PI (provider independent) address space is the way to go. The address space will always be yours, and you can take it with you.
For residential users who like to know where their devices are... that one is a bit tougher. Really DNS is your friend. How often do you change ISPs anyway?
And from the RIPE address plan manual
"So a /48 should be used when there is any doubt whether a /56 is sufficient in the long run. ISPs /48–even in
get much leeway in determining the prefix size they give to their customers up to
the case of home users"
I would say there is always a doubt that a /56 may be insufficient. A /56 only allows for 256 networks.
Microsoft http://www.networkworld.com/ar...