4 ms·

18 quintillion . . . why?

InclinedPlane 16y ago
Effectively unlimited is better than limited. It turns out that the difficulty of supporting an 18 quintillion large address space is not significantly different from the difficulty of supporting a 4 billion large address space. NTFS5, for example, can support file sizes up to 18 quintillion bytes, and it's a decade old.
- pyre 16y ago
  > NTFS5, for example, can support file sizes up to 18 quintillion > bytes, and it's a decade old. IIRC, NTFS has loads of advanced features, but Microsoft is slowing "turning them on" by adding support for them in Windows. I'm sure Windows would blow up if you got anywhere near that limit.
shkb 16y ago
Well we ran out of 4 billion, and much of the world isn't really online yet. Then factor in having a laptop, a smartphone, gaming console, HPTC, etc. There's more servers and infrastructure too. It's not that much overkill either, only 4x more bits.
- jmillikin 16y ago
  The article states that 18 quintillion is being allocated to each customer, which does seem a bit excessive.
  - bodyfour 16y ago
    (recycling an post I just made on the same topic on /.) In short: 1. an ISP would be allocated a /32. This sure sounds like a lot (2^96 addresses!) but it only is the same burden as a single address in IPv4! Even if there were millions of ISPs, this would be no burden on the address space at all. We'd lose the ability to route before we'd exhaust the addresses. 2. Then an ISP can assign a /64 to customers. Again, this sounds like a bit much, but its a tiny fraction of their space. Even if the ISP had a billion customers like this they would be completely fine. 3. This leaves a huge amount of space available for the customer to segment however they please. 4. In the simplest case you'd use a single network with 48-bits taken up by the ethernet MAC address (removing the need to do any address assignment!) Alternatively you could segment the space a bit more and run a transnational corporation's network inside the /64. Up to you! 2^128 is a really REALLY big number. No reason not to spread out a little bit.
    - swolchok 16y ago
      Yeah, and 2^32 computers should be enough for anybody. ;)
      - pyre 16y ago
        I know you're joking, but I think the point is that the numbers are so large that we really don't have the tech right now to exhaust the space. I remember hearing in the past that 2^128 is close to or more than the number of molecules (or was it atoms?) in the known universe. We're not at the point where we can assign an address to every atom and make use of it... yet at least.
SoftwareMaven 16y ago
Not sure why the down-votes. I was surprised that they were allocating 64 bits of address space to each user as well, but I don't know the details of IPv6. I think I'd be OK with 16 bits myself. :)
drdaeman 16y ago
Because that's the minimum block size (/64) for several IPv6 features - mostly important the stateless address configuration - to work. I'd thought of explaining with my own words, but found very good and detailed explanation here: http://serverfault.com/questions/182881/why-add-an-ipv6-address-as-64/218432#218432 http://serverfault.com/questions/182881/why-add-an-ipv6-addr..., so I'd just quote. --- begin quote -- The IPv6 addressing architecture (RFC4291) requires all unicast addresses always have a prefix length of 64 bits. Using something other than a /64 will break a number of IPv6 features such as neighbor discovery, secure neighbor discovery (SEND), privacy extensions, mobile IPv6, embedded-RP (multicast), etc. The current IPv6 address assignment guidelines call for allocating a /64 only when it is known that one and only one subnet is needed, otherwise a /56 should be allocated to small sites that are "expected to need only a few subnets over the next 5 years", or a /48 for larger sites. (ARIN policy 6.5.4.1) --- end quote --- This is why all those just-bit-counting "number of atoms in universe" and "79 octillion times" comparisons are wrong. Actual comparison should compare single IPv4 address with something ranging from /64 to /48 IPv6 allocation - /64 is a minimum per L2 broadcast segment - so it's a minimum per-customer allocation (unless ISP is really greedy). Also, I don't know of reasons, but RIPE NCC themselves uses /32 IPv6 block with about /21 IPv4 one comparison to calculate LIR billing score - http://ripe.net/membership/billing/calculation.html http://ripe.net/membership/billing/calculation.html
j_baker 16y ago
Fun fact: the square root of the number of available IPv6 addresses is 18 quintillion. Therefore, there are 18 quintillion blocks of 18 quintillion addresses. We're not going to run out anytime soon.
dfox 16y ago
Actually, /64 seems too small for customer allocation. What if customer wants to have more than one network?

Effectively unlimited is better than limited. It turns out that the difficulty of supporting an 18 quintillion large address space is not significantly different from the difficulty of supporting a 4 billion large address space. NTFS5, for example, can support file sizes up to 18 quintillion bytes, and it's a decade old.

pyre 16y ago
> NTFS5, for example, can support file sizes up to 18 quintillion > bytes, and it's a decade old. IIRC, NTFS has loads of advanced features, but Microsoft is slowing "turning them on" by adding support for them in Windows. I'm sure Windows would blow up if you got anywhere near that limit.

shkb 16y ago

Well we ran out of 4 billion, and much of the world isn't really online yet. Then factor in having a laptop, a smartphone, gaming console, HPTC, etc. There's more servers and infrastructure too. It's not that much overkill either, only 4x more bits.

jmillikin 16y ago
The article states that 18 quintillion is being allocated to each customer, which does seem a bit excessive.
- bodyfour 16y ago
  (recycling an post I just made on the same topic on /.) In short: 1. an ISP would be allocated a /32. This sure sounds like a lot (2^96 addresses!) but it only is the same burden as a single address in IPv4! Even if there were millions of ISPs, this would be no burden on the address space at all. We'd lose the ability to route before we'd exhaust the addresses. 2. Then an ISP can assign a /64 to customers. Again, this sounds like a bit much, but its a tiny fraction of their space. Even if the ISP had a billion customers like this they would be completely fine. 3. This leaves a huge amount of space available for the customer to segment however they please. 4. In the simplest case you'd use a single network with 48-bits taken up by the ethernet MAC address (removing the need to do any address assignment!) Alternatively you could segment the space a bit more and run a transnational corporation's network inside the /64. Up to you! 2^128 is a really REALLY big number. No reason not to spread out a little bit.
  - swolchok 16y ago
    Yeah, and 2^32 computers should be enough for anybody. ;)
    - pyre 16y ago
      I know you're joking, but I think the point is that the numbers are so large that we really don't have the tech right now to exhaust the space. I remember hearing in the past that 2^128 is close to or more than the number of molecules (or was it atoms?) in the known universe. We're not at the point where we can assign an address to every atom and make use of it... yet at least.

SoftwareMaven 16y ago

Not sure why the down-votes. I was surprised that they were allocating 64 bits of address space to each user as well, but I don't know the details of IPv6. I think I'd be OK with 16 bits myself. :)

drdaeman 16y ago

Because that's the minimum block size (/64) for several IPv6 features - mostly important the stateless address configuration - to work. I'd thought of explaining with my own words, but found very good and detailed explanation here: http://serverfault.com/questions/182881/why-add-an-ipv6-address-as-64/218432#218432 http://serverfault.com/questions/182881/why-add-an-ipv6-addr..., so I'd just quote. --- begin quote -- The IPv6 addressing architecture (RFC4291) requires all unicast addresses always have a prefix length of 64 bits. Using something other than a /64 will break a number of IPv6 features such as neighbor discovery, secure neighbor discovery (SEND), privacy extensions, mobile IPv6, embedded-RP (multicast), etc. The current IPv6 address assignment guidelines call for allocating a /64 only when it is known that one and only one subnet is needed, otherwise a /56 should be allocated to small sites that are "expected to need only a few subnets over the next 5 years", or a /48 for larger sites. (ARIN policy 6.5.4.1) --- end quote --- This is why all those just-bit-counting "number of atoms in universe" and "79 octillion times" comparisons are wrong. Actual comparison should compare single IPv4 address with something ranging from /64 to /48 IPv6 allocation - /64 is a minimum per L2 broadcast segment - so it's a minimum per-customer allocation (unless ISP is really greedy). Also, I don't know of reasons, but RIPE NCC themselves uses /32 IPv6 block with about /21 IPv4 one comparison to calculate LIR billing score - http://ripe.net/membership/billing/calculation.html http://ripe.net/membership/billing/calculation.html

j_baker 16y ago

Fun fact: the square root of the number of available IPv6 addresses is 18 quintillion. Therefore, there are 18 quintillion blocks of 18 quintillion addresses. We're not going to run out anytime soon.

dfox 16y ago

Actually, /64 seems too small for customer allocation. What if customer wants to have more than one network?