On Fri, Oct 23, 2015 at 07:06:26AM -0700, S Moonesamy wrote:
>
> There is a news article at http://www.lexpress.mu/article/251452/tic-un-point-dechange-internet-local-pour-reduire-couts-connexion
That has some very funny comments! :-)
> that news article. From what you wrote above, it looks like even if the IX
> works it does not mean that the internet (in Mauritius) will be faster.
It depends. How do you connect to the internet? (does *your* isp peer?) What do you consider
"faster"? (gaming latency? Speed/throughput of a Linux .iso download?
Perceived page load time on your favourite new site?) What to you is
"the Internet"? Is it Facebook? Is it your local internet banking? Is it
your inter-company business emails? Is it Skype calls to your aunty in
Australia?
The article is obviously making generalised sweeping statements. It
should be obvious from the above questions that of course an IXP located
anywhere cannot and will not magically make all of "the internet" faster
for every user in that locality all of the time.
However, if you consider "The Internet" as the sum of all connections
that all the billions of people around the globe use it for.. And if you
then consider that for a number of people, certain data they want to
access, they can access quicker than yesterday.. Then as a whole, on
aggregate, wouldn't you agree "The Internet" is a little bit faster
today?
> In my opinion, there are different objectives here; it is in the interest of
You talk about objectives. That is an aim, an purpose, a target, a goal.
Something in the future to achieve..
> According to the above the MIXP is seeing as much traffic as a user on My.T
> 20M (without FUP) would generate.
.. and yet you focus on measurements over a few weeks or months past.
Think about a 5 or 10 year plan. Where do you want your local internet
to be then? Are there a whole number of barriers to achieving Internet
nirvana? Does the existence of an IXP lower or remove any of those
barriers at all? Could it help enable businesses, start-ups or access to
education in the future?
It might. It might not. But *could* it?
> Let's not call it a failure; how about calling it a white elephant? :-)
>
You have your view and can call it whatever you like. I call it a very
good start.
As a couple final comments:
a) You should read this in full:
http://drpeering.net/a/Internet-Peering-Playbook/Second-Edition/Source/HTML_IPP/chapters/ch13-0-1-Value-of-an-IXP/ch13-0-1-Value-of-an-IXP.html
Then read it over again and understand that measuring value of an IXP is
all economics and very little else.
Then consider that for the time being at least the opex cost of a
network to participate at the MIXP is pretty much zero.
Then read the above again and think about why it's worthwhile for
network operators to make the once off capex investment to get
connected, no matter how small the total traffic they might pass
initially.
You may then ask again about the benefit to a home user... Well, it's
simple if we stop thinking about how the next youtube video loads and
start thinking about the big picture of the economy and the business of
network, internet service and hosting providers.
If networks can make cost savings by exchanging traffic, does this
provide a way for smaller networks to remain competitive in a tough
market? Do more smaller networks existing foster competition? Does
competition keep service levels going up and costs down?
b) A final little exercise:
Use your favourite DNS tool (dig, nslookup etc) to check the name
servers for your ccTLD .MU - then do a traceroute to each one and see
what happens with one of them.
- Daniel
Received on Fri Oct 23 2015 - 21:06:52 PST