Velocity Reviews > IPv4 Addresses: 5% Left

# IPv4 Addresses: 5% Left

victor
Guest
Posts: n/a

 10-21-2010
On 21/10/2010 5:41 p.m., Matty F wrote:
> On Oct 21, 12:48 pm, Lawrence D'Oliveiro<l...@geek-
> central.gen.new_zealand> wrote:
>> In message
>> <(E-Mail Removed)>, Matty F
>> wrote:
>>
>>> Why didn't they use IPv6 in the first place?

>>
>> We’re talking pre-PC era, back when computers, at least ones powerful
>> enought to be worth networking, were hefty boxes that had to live in their
>> own special rooms. Nobody ever thought we’d have a billion computers like
>> that. And indeed, we never did.

>
> I invented a principle (but somebody else may have thought of it
> first) that when you are designing something, assume that it is going
> to succeed, and design accordingly to allow it to be scaled up.

Its a principle invented by a Mr I M Smug known as "hindsight".

John Little
Guest
Posts: n/a

 10-22-2010
On Oct 20, 9:45*pm, Lawrence D'Oliveiro <l...@geek-
central.gen.new_zealand> wrote:

> Given that IPv6 has room for roughly the same number of addresses as the
> estimated number of atoms in the entire observable Universe...

That's an exaggeration. 2^128 is about 3 x 10^38, cricket ground for
number of atoms is 10^80.
(See http://en.wikipedia.org/wiki/Observa...atter_content).
Just 41˝ orders of magnitude...

Also, subnets are always 64 bits in IPv6, drastically reducing
utilization of the address space.

Regards, John

John Little
Guest
Posts: n/a

 10-22-2010
Lawrence said:
> >>> math.log(2 ** 128, 3)

>
> 80.75900845714655

What's that 3 for? What relevance is a log to base of 3?

>>> math.log(2 ** 128, 10)

38.531839444989586

Matty F
Guest
Posts: n/a

 10-22-2010
On Oct 22, 10:19 am, victor <(E-Mail Removed)> wrote:
> On 21/10/2010 5:41 p.m., Matty F wrote:
>
>
>
> > On Oct 21, 12:48 pm, Lawrence D'Oliveiro<l...@geek-
> > central.gen.new_zealand> wrote:
> >> In message
> >> <(E-Mail Removed)>, Matty F
> >> wrote:

>
> >>> Why didn't they use IPv6 in the first place?

>
> >> We’re talking pre-PC era, back when computers, at least ones powerful
> >> enought to be worth networking, were hefty boxes that had to live in their
> >> own special rooms. Nobody ever thought we’d have a billion computers like
> >> that. And indeed, we never did.

>
> > I invented a principle (but somebody else may have thought of it
> > first) that when you are designing something, assume that it is going
> > to succeed, and design accordingly to allow it to be scaled up.

>
> Its a principle invented by a Mr I M Smug known as "hindsight".

It's in Systems Analysis 101, if a field is likely to expand in size,
have a field length on it.

Lawrence D'Oliveiro
Guest
Posts: n/a

 10-22-2010
In message
<(E-Mail Removed)>, John
Little wrote:

> Also, subnets are always 64 bits in IPv6, drastically reducing
> utilization of the address space.

No theyâ€™re not. They learned their lesson from the â€śclassâ€ť restrictions of
IPv4.

Lawrence D'Oliveiro
Guest
Posts: n/a

 10-22-2010
In message
<(E-Mail Removed)>, Matty F
wrote:

> It's in Systems Analysis 101, if a field is likely to expand in size,
> have a field length on it.

Does that mean you should have a length field for the field length field,
just in case?

Matty F
Guest
Posts: n/a

 10-22-2010
On Oct 23, 11:51 am, Lawrence D'Oliveiro <l...@geek-
central.gen.new_zealand> wrote:
> In message
> <(E-Mail Removed)>, Matty F
> wrote:
>
> > It's in Systems Analysis 101, if a field is likely to expand in size,
> > have a field length on it.

>
> Does that mean you should have a length field for the field length field,
> just in case?

Yes, if you are going to have fields longer than 255 bytes. I never
saw fields longer than that in commercial systems. Longer fields could
probably be broken up into smaller fields of max 255 bytes.

Ted
Guest
Posts: n/a

 10-22-2010
On Fri, 22 Oct 2010 16:01:13 -0700 (PDT), Matty F
<(E-Mail Removed)> wrote:

>On Oct 23, 11:51 am, Lawrence D'Oliveiro <l...@geek-
>central.gen.new_zealand> wrote:
>> In message
>> <(E-Mail Removed)>, Matty F
>> wrote:
>>
>> > It's in Systems Analysis 101, if a field is likely to expand in size,
>> > have a field length on it.

>>
>> Does that mean you should have a length field for the field length field,
>> just in case?

>
>Yes, if you are going to have fields longer than 255 bytes. I never
>saw fields longer than that in commercial systems. Longer fields could
>probably be broken up into smaller fields of max 255 bytes.

Or have variable length fields. Something like UTF-8, but endless.

Matty F
Guest
Posts: n/a

 10-22-2010
On Oct 21, 7:30 pm, Lawrence D'Oliveiro <l...@geek-
central.gen.new_zealand> wrote:
> In message
> <(E-Mail Removed)>, Matty F
> wrote:
>
>
>
> > On Oct 21, 12:48 pm, Lawrence D'Oliveiro
> > <(E-Mail Removed)_zealand> wrote:

>
> >> In message
> >> <(E-Mail Removed)>, Matty
> >> F wrote:

>
> >>> Why didn't they use IPv6 in the first place?

>
> >> We’re talking pre-PC era, back when computers, at least ones powerful
> >> enought to be worth networking, were hefty boxes that had to live in
> >> their own special rooms. Nobody ever thought we’d have a billion
> >> computers like that. And indeed, we never did.

>
> > I invented a principle (but somebody else may have thought of it
> > first) that when you are designing something, assume that it is going
> > to succeed, and design accordingly to allow it to be scaled up.

>
> So how scalable were your systems? Could they network a billion machines?

I suppose there always has to be a limit somewhere.
In the early 1980s we allowed the addressing for our database to reach
8 Terabytes.This was in the days when disk drives for PCs were seldom
bigger than 256 Megabytes. I never imagined that a commercial system
would need more than 8 TB. So I allowed for a size increase of 30,000
times over a 30 year period, and that was not enough. However it was a
simple conversion to copy the database with another byte of
addressing. The database did have a 1 byte length for the length of
the address field. So the database could never be bigger than the
number contained in a 255 byte binary field. I believe that is enough.

I designed and implemented a three dimensional spreadsheet whch could
have 2,000,000,000 rows, columns and another dimension. Each cell
could contain a query that could examine a simple file of up to
2,000,000,000 records.
Numbers could have almost any editing desired, up to 64 characters in
length. Not bad for the1980s. It's still larger than Excel.

Murray Symon
Guest
Posts: n/a

 10-23-2010
Lawrence D'Oliveiro wrote:

> In message
> <(E-Mail Removed)>, Matty
> F wrote:
>
>> It's in Systems Analysis 101, if a field is likely to expand in size,
>> have a field length on it.

>
> Does that mean you should have a length field for the field length field,
> just in case?

DER has a scheme that handles that pretty well.
It is TLV with a variable Length field.