On Thu, 11 Aug 2011, Rob Evans wrote:
> > Interesting. We successfully argued that a /48 was insufficient to
> > avoid a fragmented internal allocation strategy, and I believe
> > Cambridge have subsequently done the same - both of us now have /44s,
> > but perhaps we should have pushed for more :-)
>
> Should 64,000 subnets (a /48) or 1,0000,000 [1,000,000] subnets (a /44)
> not be enough :), ...
But as with all IPv6, the raw numbers are just a red herring. Of course
the number of subnets is more than enough, as is the number of hosts
possibly in each /64 of subnet. Just one of those subnets would be big
enough for all our hosts of course, but the point is to encourage
aggregation and delegation, those 64,000 (or whatever) subnets are
sparsely populated, but suitable blocks sub-assigned to OUs etc.
IPv6 was promoted as "huge address space, enough for millions of addresses
per person blah blah", but it's not quite so huge when there is such great
emphasis on carving it up in big blocks, and sparsely allocating for
"futureproofing". Much is 'wasted'.
I can see a case for a very devolved collegiate environment like Ox or Cam
having a bigger allocation. Probably /48 is fine for most of the rest of
us. And I think in the JANET allocations the allocations are sparse to
allow expansion of the bits for sites who find they need more in the
future?
Jethro.
. . . . . . . . . . . . . . . . . . . . . . . . .
Jethro R Binks, Network Manager,
Information Services Directorate, University Of Strathclyde, Glasgow, UK
The University of Strathclyde is a charitable body, registered in
Scotland, number SC015263.
|