Category Archives: Computers

ARIN reaches final /8

Posted on by

In slightly unexpected news, ARIN reached their final /8 today supposedly because of the allocation of a /10 to Akamai, which reduced the space free from just over 1.25 /8s to 1 /8s exactly.  This means that ARIN are now in ‘Phase 4’, where all address allocation requests are closely scrutinised, and requests for /15 are above are closely scrutinised by a senior member of ARIN.

I haven’t seen the actual allocation figures yet, but this really should be the starting gun for Americans everywhere to start thinking about taking their heads out of the sand and rolling out IPv6.  It’s now going to be a lot harder to get allocations out of ARIN, and it can’t be long before total exhaustion is upon us.  Geoff Huston’s meter is probably going to be wound forward by months in the morning.

LACNIC crashes through the ‘final /8’ barrier in style

Posted on by

LACNIC crashed through the ‘final /8’ barrier in style yesterday by allocating a smashing 2,368,512 IPv4 addresses in just one day (0.141 /8s).  This brings the total left to around 0.87 /8s as of today, and this really does signficantly reduce the time needed to reach exhaustion (which in LACNIC’s case will be when they reach the final /10, or 0.25 /8s).

Most of the addresses allocated yesterday went to Brazil, the biggest block of 1,048,576 (/12) addresses going to Tim Celular S.A. in Brazil.  Other notable blocks include 2 x 262,144 addresses (/14)  and 6 x 131,072 (/15) going also to Tim Celular at the same time.  All the others are blocks of 1,024 and 2,048 (/22 and /21) so are relatively insignificant.

All blocks apart from two were allocated to Brazil, apart from one /22 which went to Colombia and another /22 went to Costa Rica.

It won’t take many more allocations like this to bring exhaustion very very close now.  Are there any other ISPs or mobile phone companies need more addresses in the near future, I wonder?

IPv4 depletion at LACNIC

Posted on by

As of today, LACNIC (the IP address registry for Latin America and the Caribbean) are down to 1.05 /8s remaining, which means that it is likely in the next day or two to have reached the final /8 that they have.  LACNIC have slightly different rules to some of the other registries – RIPE and APNIC, for example, considered their stock of IPv4 addresses to be exhausted when they reached the final /8 (16,777,216 addresses) , whereas ARIN are taking a different approach and are basically reserving hardly any, so it truly is a “race to the bottom” for that region although the requirements for obtaining a block when they reach the final /8 are much increased.  LACNIC consider their addresses to be exhausted when they reach 4,194,304 addresses (/10) left.

Yesterday there were 524,288 addresses (/13) allocated to Telemar Norte Leste S.A. in Brazil, and a further 393,216 addresses split as two blocks of 262,144 (/14) and 131,072 (/15) addresses to Telefonica de Argentina.  If this rate of allocation continues (i.e. 0.05 /8s per day), LACNIC will be out of addresses in 16 working days from now.  I suspect it won’t be depleted that quickly, but still a sobering thought, if you still have your head in the sand over IPv4 exhaustion and you’re located in the LACNIC region.

The End of Year IPv4 Report

Posted on by

So, here we are at the end of 2013 and let’s see where we are.  So far, APNIC and RIPE have ‘run out’, and the next two registries likely to be depleted in the next year or two are ARIN and LACNIC.  Until recently, it was likely that ARIN would run out first, and then ARIN and LACNIC were pretty much neck-and-neck for a while, but in recent months the demand for IPv4 in the LACNIC region has been accelerating quite a lot, and we’re now at the point (as of today) that LACNIC are now depleting their resources quite a bit faster than ARIN are.  Geoff Huston has the figures as of this morning at 1.5 /8s for ARIN and 1.3317 /8s for LACNIC, a difference of approximately 6 weeks.  If LACNIC continue this rate of depletion I suspect they will be down to the final /10 (as ARIN and LACNIC have different rules to RIPE, who reserved a whole /8) a fair bit sooner than the beginning of 2015.  As to why LACNIC has shown such a spurt in recent weeks, that’s anyone’s guess, but the next few months could be very interesting if current trends continue.  Just as well Comcast in the US is well on the way to rolling out IPv6 to 100% of their customer base then!

Happy new year!

ARIN reaches Phase 3, with only two /8s left

Posted on by

Today the American Registry for Internet Numbers (ARIN) reached the next milestone in its IPv4 exhaustion journey.  There are now only two /8s remaining left, and one of those is the “last /8” which both APNIC and RIPE are already down to.  By current estimations (and particularly Geoff Huston’s), we’re probably about a year away from reaching the final /8 at ARIN.  This, of course, depends on whether the current depletion rate is maintained (which is what happened in the RIPE region), or whether there is an APNIC-style “run on the bank” and panic sets in.  Either way, ARIN’s page on the subject is interesting, as it breaks down the number of contiguous blocks available between /8 and /24.  If any of the big ISPs want a really block, they’d better get in quick, since at the time of writing there’s only one /8 left, two /10s, one /11, two /12s, one /13 and three /14s.

Interesting times ahead, and it will be fascinating to see which way it goes.  Especially as ARIN does not appear to have adopted a “final /8” policy as APNIC and RIPE have done, so it really is a dog-eat-dog race to the finish, pretty much.

New OpenPGP Key

Posted on by

Just to let you all know I have a new OpenPGP key, which is 4096 bit RSA.  Available from all the usual PGP key sites, or you can download directly from the Contact page should you prefer, where you’ll find the fingerprint details.

Why you should never trust companies who make promises containing the word “forever”, part 4,294,967,296

Posted on by

Google recently announced that they would be turning off their Google Talk service and replacing it with Google Hangouts instead.  Whilst normally I wouldn’t care at all about this, and to a certain extent I still don’t, by doing this they have gone back on something that Google said some time ago about using open protocols and the word “forever”.  Not that this was in any way a surprise to me that they’d do something like this, but it may be to some people, since their whole business model relies on selling your data which you provide to them for nothing, so if they’re not making money on a certain feature, I’d imagine they’d have no qualms about pulling it to suit themselves, rather than you.

As you may know, Google Talk was based on a protocol called XMPP (or Jabber as it was formerly known), which is an open standard for instant messaging, defined by RFC 6120-6122 et al.  The advantage of using XMPP, in addition to the fact it is an open standard, is that there are a huge variety of XMPP clients and servers you can use, whether on PCs, Macs, Linux, or tablets and phones, and they all interoperate fine, in the main.

Google’s decision to stop federation of XMPP between “third-party” (to Google) servers and their own basically means they are limiting the use of their servers to Google account holders only.  I expect that in due course, Google Talk itself will be retired fully so that their own official client will no longer work.  (The Android app has already been replaced with “Google Hangout”.)  This is not unlike Facebook Chat, which also uses XMPP, but at least they have been clear from day one that they will not be federating with anyone else and have always operated a “closed” server.

It’s a stark reminder that when a company promises “free”, there’s usually a catch, and if they promise “forever”, that means “until we feel like discontinuing it”.  People who were relying on both of those terms meaning what they say they do, have just received a nasty shock.

If you were using Google Talk to communicate with my XMPP server, then you will now need to find another XMPP account to use instead.  Thankfully, there are many different servers around the world who offer such accounts, one of the best known being jabber.org but many other sites are available too, and this link has a list of some of them.  You can of course, if you are able to, run your own XMPP server using one of many open-source or commercial XMPP servers available.

Once you’ve registered an account, if you need an XMPP client, there are many to choose from, and many are open source.  Popular ones available for Windows, Mac and Linux include Psi, Pidgin, Adium, and many more.  For Android users, Xabber is available for Android from the Google Play store, and there are also plenty for other types of smartphones (though I have no direct experience of them).  Some of these clients also can do voice calls now, which again due to a choice made by Google, was never available between third-party servers and Google Talk, only between their own customers.

At least Google’s decision hasn’t killed XMPP, the protocol, which is still an open standard, and still has millions of users worldwide, I’d encourage everyone to continue to use a standard which isn’t going to go away because one particular (large) company says so.

(And, in case you’re wondering, my Google Hangouts app on my Android phone is still disabled, just as the Google Talk one was before it.)

BT Retail quietly trials CGNAT on some of their customer connections

Posted on by

According to this news story on ISPreview.co.uk, BT Retail are quietly trialling Carrier Grade NAT on some of their customer connections,  This is the first of the ‘Big 5’ that are known to be doing so, not long after the Plusnet (also owned by BT) trial took place.  To me, this suggests that we are now at the point where even the big ISPs no longer have the IPv4 addresses to allocate to customers, and the article suggests that up nine other customers will be sharing a single address, if you have been placed on this trial.

The worry here is that all of the Big 5 will just use this as an excuse to delay IPv6 adoption further, when what we really need is mass adoption of IPv6 folllowed by, if and only if necessary, CGNAT for all the “legacy” applications still left on the Internet.

BT’s official FAQ on the matter can be found here.