[lugm.org] Why IPv6? [was: Announcing a Mauritian tech forum]

[Nishal Goburdhan]

On 18 Dec 2013, at 6:28 PM, selven <pcthegreat at gmail.com> wrote:

> > the short answer is that you live and work in an environment where the basis for your connection (IPv4) is now depleted.  there is *no* other sustainable solution for the future other than IPv6, so if you plan to continue to live and work in this environment, when do you plan to develop the skills to do this?  when you absolutely need it in a pinch, or now (or, 5years ago) so that you're prepared for this transition.
> 
> True. But am a coder and sometimes into architecture. Shifting v4 to v6 is just a script away. The skills needed, that has always been read the manual read the rfc mix wiv instinct. So shifting does not seem a big deal hence why i believe without the pressure to shift to v6 i wnt :) money talks. Business owners decides that :) and i do only what am paid to do during work time. Yes i do advise, but i will never 'convince'.

writing *good* code, also means that you have an implicit understanding of How Things Work - that is more than "just a bit-shift" for IPv6, as there are fundamental changes to the way that some things work in IPv6.  even the most skilled coder requires time, and experience to hone the skills to master these?

when is that time?    this was posted to the nanog mailing list [1] earlier this week.  a rough indication of IPv6 traffic from these operators:
Google Fiber 70%
Virginia Tech 61%
Verizon Wireless 40%
Comcast 20%
AT&T Wireline 15%
T-Mobile US 6%

one local university in ZA [2] has gone on record saying that once they'd enabled IPv6 end-to-end across their network, more than 50% of their network traffic was IPv6. 

as a coder, you're should also be more than agreeable, that a return to the end-to-end principle of the internet, would make your life much simpler!   :-)
there's a quote somewhere (my google-fu is failing me right now) from the designers of skype, explaining that if they (skype) didn't have to counter for the complexity of NAPT, the skype binary would be orders of magnitude smaller.

so - want to write cleaner code - is IPv6 the way to do that ?


> So isnt there a consortium to convince the world to shift?

if there wasn't a consortium to make the world shift to IPv4 first, and then later, classless addressing, why would you think that there would be such a cabal now?  people do it, because they see the need to  (ok, in the case of classless addressing, it was different...)

you can choose - either develop the skills over time (while you can).  
so when it comes down to the crunch, you won't need an expensive consultant to do this for you overnight.
it's your MUR, so, really, it's your call.  


> But some just make things works with just a linux or bsd box. The pressure to shift to v6 is not yet there. Because it doesnot bring any added profit yet. And like i said earlier, shifting now or shifting later is a matter of at most one hour downtime (worst case scenario). 

i'm willing to bet, that once you *do* decide to move to IPv6, it will *definitely* take you more than an hour.  at the very least, planning, and implementing, definitely will.  when did you plan on developing those skills - within that same 1hr gap?  
i've been running IPv6 on networks and services for about 10years on/off now.  i still get stuff wrong occasionally.  or, my operator/service provider does, and that 1hr time-window that you're talking about, suddenly extends to ... much longer periods.

then again, perhaps when everyone else starts using IPv6, you'll all be a lot smarter than i am ;-)


> > you can get lots more info here:  http://www.internetsociety.org/deploy360/ipv6/all-ipv6-resources/
> 
> Lots of papers not enough convincing of business owners. Can iss fix my latency issue because my isp has bad knowledge of routes or doesnt peer wisely? Can iss convince the mauritian society to shift to v6 to ease my life (yes v6 everywhere wud had eased my life but if it is not every it does not). The answer is probably no, which is why i do not believe there is anything worthwhile in moving now

i think you're confusing many different things here.  there is a clear, and undeniable need for network operators to learn to build stable, efficient, and scalable networks - i know this better than you might imagine :-)   

however, building that, is an orthogonal question to why you should get involved with IPv6.  true - business owners are not convinced.  but, let's be honest about that - the blame there is largely on the engineering people.  because, as a good engineering resource, your job is to forcecast what it will take to operate your environment into the next 3-5yr cycle.  if you aren't taking into account that you might not be able to get IPv4 anymore, or that the cost to to this will increase exponentially, then you're doing your organisation a disservice. 

at the last ISP that i worked for, my "business owners" didn't care a fig what an IP address was.  what they cared about, was that if they needed us to deploy something in $time_frame, and with $predictable_cost, that is could be done.   

so, my question to you is:  how are you, as a technical resource, going to keep $cost = predictable, when you realise that:
* you either can't get IPv4
* it comes at a significantly higher cost than was forecast, because you're now getting it from an IP address broker, and not a RIR
* your clients expect this thing call IPv6, which your team isn't trained on
... etc? 

it sounds to be that you understand that there is no likely future, other than IPv6.  (which, is a good thing :-))  but you're not sure when you should start learning/planing/investing in these skills.  so i'll ask a different question:  if you know, that it's going to rain, do you carry an umbrella with you from the start, taking your time to pick the colour/vendor, or wait for it to start raining before buying one - at which point you're limited to what the immediate vendor has to offer you, and, since you need it *there and then* you're more than likely willing to pay whatever they ask you for it?   :-)

i teach a lot of network related stuff;  if you ever sit in any of my classes, you'll hear me say, over and over again:  "the key is success is planning..."


> > > Mauritian hosting easy to say but how do you provide cheap hosting given the crappy connection versus price we have here?
> >
> > economists talk about economies of scale that are needed to drive down prices.
> > how do you *ever* plan to drive down pricing to what you're willing to accept as "decent", if you don't start building scale ?
> 
> Economist also killed the barter system so as to allowed those who cant produce to live like bacterias off others.

actually no, that's not true  :-)  
economic theory says:  institutions exist because of transaction costs.  
ie.  it's easier (and cheaper) for you to do something (write code), and trade services with me (network tech) allowing us to improve skills and develope even greater proficiency in each, thus serving ourselves, and each other better.   that system has been .... re-written and (ab)used for other things, many times over the past 2-300 years.... 


> How do you start anything hosting wise even local when you have stupid isps who if u want to ping an adbn box in the same country as you makes your packet travel to london (linx maybe) then back to mauritius. This wud be fighting against the current.

honestly, i think that the answer here is easier than you think.  and maybe we can discuss this further on the IX thread that KN is going to start on binary.mu  ;-)


> > if you continue to invest in the $3 SaaS $colo_provided_vm, you're never going to grow your local business sectors.  well, that is until your regulator realises that, as a country you're spending XX-million hosting outside the country, and in an effort to improve the BoP, they pass a law that says all mauritian content *must* be hosted in-country.   (really, you think i'm dreaming?  think again!!)
> 
> If i want to make a business i shud first of all think about filling my pockets, making the country happy is secondary. We are not in a perfect world where you can do this.

that's a very, very short-term view on business.  the largest broadband ISP in ZA today, didn't turn a profit for about 10years.  their strategy was to build market share, perform acquisitions, etc.  today, they're the 600lb gorilla in the room.
so, sure, if your business strategy is setup for a year, then you want to line your pockets.  if you have a more holistic plan (no, it doesn't need to be 10yrs...that was just an example)  perhaps you would look at approaching things differently.


> > today, we have lots of colo centres in ZA;  local hosting is a fraction of what it used to be a decade ago.
> See my comment about local boxes connecting to each other might reach over 280ms rtt.

seriously?  what makes your environment special? 
where do you think ZA was back in 1996?  :-)
how do you think we got better? 

if you're identifying "investing in local peering" as a key component to growing hosting, then, awesome.  invest time/energy/infrastructure/skills/legislation (in whatever manner you can) to make that happen.  


> > it's still more expensive than a VM in SFO, but, there's sufficient realisation that there is true value in keeping content in country.
> 
> Keeping content in country if it raises your expenses makes customers unhappy and makes  u earn lesser. Every has to eat. I do not make my country happy when i know corruption causes incompetent ppl to work at positions that cud have helped us gain better service. Am not a moron to enrich the country so as ppl steal from it. I prefer to salvage and sabotage it instead so as bacterias and viruses do not see anything to steal and leave it alone *cough* *politicians*

see my earlier comment about bandwidth pricing differentials. 
you really, really, need to step back from your existing scenarios, and look objectively at possible solutions.

i have no comment on your "corruption" issues - that exists everywhere.  just look at the ZAR206mil that was apparently spent on our president's home.  that doesn't stop me paying my taxes, and still, continuing to work for what i believe is the betterment of my community.  ymmv.


> > (for a start, it actually *does* make it a lot tougher for the NSA to get to... :-p)
> > how are you - as the assumed informed crowd - promoting this want/need/desire to keep things in-country?
> > because - honestly - if you don't sit up and promote these ideas, whom do you expect to?
> > do you active choose *not* to purchase from a non-local site?
> > and then *tell* the operator why you're boycotting him?
> 
> You cant boycott the isp with the monopoly.


....you make it sound like there is no other choice.  i know at least three ISPs that provide their own transmission services in and out of mauritius.
so...with respect, you're wrong, dude  :-)


> Sbm is still running. And it is still outsourcing it is normal stuffs.

ah, i didn't plan on taking down SBM.  that's not rational.  and i had a small social circle in MU, so i could do less to influence people.
but, i chose to vote with my wallet.  and i educated at least 4 other people on why that was bad.  
and, i believe, at least one of those people, has passed on the same message.

perhaps if more informed mauritians did the same...well, ripples in a pond...
my challenge to you is this: 
* what are the things that matter to you?   what are the things that you want changed?   and what are you - as the smarter, more connected, tech-know-hows doing, to educate, and influence for that change ?

i did not ever say that your results would be immediate, or that they would be easy to achieve.  


> > on my last trip to MU (circa May 2013) i spoke to an ISP that had recently put in a direct interconnection, between ZA and MU.
> 
> That is adbn. Ex nomad. It is not for common users. And i called to get a business line. I called thrice and the sales person was still lunching for 3 hrs.
> I do not call more than 3 times.

i try not to mention names, but now that it's out there.  yes, it's nomad.
you're childish if you judge the potential and actions of an entire company based on one person.  as i said, i visited them, and they quoted me later than evening.  perhaps i was lucky, but, in my experience, if something matters enough to you, you'd make an effort to get it done...


> > i just checked - it's still in place.  so, right now, that's roughly 60ms between JNB and ebene.
> 
> 54ms to be exact
> Try mt to an adbn box now.

again, see my comment about fixing local peering above :-)


> > right now, both their regular competitors have roughly 400ms back to MU from my home in JNB.
> > they (progressive ISP) were looking for ideas on how to resell this capacity.
> > out of curiousity, i asked them for regular pricing;  they were on par 15% cheaper than two other larger ISPs that i know of.
> > so.  cheaper, and faster.  mm...can you honestly tell me that you can't think of at least seven different business ideas here, none of which involve hosting in MU?
> 
> Told you i already went through this. Keshwarsing can attest :)
> 
> > (and really, if you can't, then let me know if you have oodles of venture capital, because i certainly can !!)
> >
> > how do you eat an elephant  ?
> 
> With friends. But why share the elephant if you hunted it alone?

...because you're never going to get through all the meat if you do this alone, before the meat spoils.   (yuck, as a vegetarian, i should have really chosen a better analogy!), 

when facing a mammoth task, it's smart to know when you either need, or can offer, assistance.  when seacom hit ZA, there was a consortium of small operators that banded together, and bought an 155mb/s circuit, so that they could get better per mb/s pricing, than the larger operators were giving them. as i said - sometimes, to get something, you just need to adjust your way of thinking! 


> > > Singapore, some ips go through malaysia some still goes through linx weird peering. That is also why am unable to use my quakelive premium account in singapore any more, latency is a b*tch.
> >
> > so, in one sentence, you complain about latency being a b***h,.....but, then collectively, largely don't see a problem with overseas hosting
> > something smells fishy here :-)
> 
> :) check local inter isp latency :) price wise it is better to go for out of country hosting.


yes. and you know what?  dell/HP/blah servers are probably always going to be cheaper in the US, than buying from a local supplier.  stop comparing costs on a per bit basis.  instead, start working out how to drive down local costing to where that price makes it more attractive/useful to you.


> > > How would you propose to solve the hosting problem without having to give up all your fortune to MT and still get any profit out of it? Setting up local exchange points and bypassing isp crazy rivalry? Who sponsors the scene?
> >
> > i have enough equipment in my home right now, to sponsor you, if you wanted to build an internet exchange point.
> 
> :) am working on a better alternative to kill isps profit at once :) coders win!
> 
> > and speaking as someone that currently manages three (3) active IXes, i can tell you that it doesn't take a lot of time and effort to run these.  but why build a new one, when there's already an IX in ebene?  better questions would be:
> 
> But peerong is decided by the isp not the exchange you cant force isps to share their routes if they do not want. 

yes and no.
in an ideal world, the ISPs peer without interference.  there are many cases of local operators being forced to peer, in order to retain existing telecoms licences;  the IX in cairo, egypt, for example, does about 1Gb/s of local egyptian traffic, because peering is enforced on them by law.  the ISPs in namibia are required to peer with each other as a condition of keeping their licences - and they do, even though there is no existing IX.
is this harming your community?  can you show it?  do you think that the community (and here, this is everyone that is paying for IP services) cares?
i've spent many hours speaking to your regulator, while i was in MU.  i think you would be pleasantly surprised as to how much they actually *do* care.  i think, nay know, that they would really value some intelligent, constructive input from civil society groups on elements of mauritian services that need improvement.  could you author a well written brief on this to them?  


> > * why isn't this IX working?
> > * what can be done to revive it?
> > * if the ISPs aren't going to do this themselves, how can we, as civil society, organise ourselves to get this done?
> 
> It is not an isp problem it is more of political order. They are happy enjoying the monopoly.

i don't read french or creole, but i thought you still had freedom of speech in mauritius....?
so your community would be "ok" hearing that they could save $$$ but are not because of inaction on the part of the ISPs (or regulator)....interesting...  :-P

(btw, taking on topical issues like this, is how mybroadband became a community force to be reckoned with in ZA...it takes time and effort of course, but there's no doubt as to the value that this can bring!)


> ---_--------------i'll reply the rest later. I need food and i need woman and i need to get high. Reply in some hours.---++++---------


as ajay rightfully said, this is now off-topic for a linux list, so if you want to continue this discussion, we can do it on binary.mu which seems to be more generic  :-)  or unicast off-list.

--n.

[1] http://mailman.nanog.org/pipermail/nanog/2013-December/062899.html
[2] http://mybroadband.co.za/news/internet/56241-shocking-ipv6-revelation-in-south-africa.html