I have no reason to post this other than its cool as hell to me for some reason
Community driven Internet Exchange. That seems right in line with how the original internet was envisioned to be ran. Just people adding to the community
@wendell @Dynamic_Gravity @TimHolus @SgtAwesomesauce It is quite cool though
Wonder if this can be done at scale with a lot of community driven IEXes around
It makes sense that they did that in some ways.
Thoughts all?
I was going to post about peering issues, but when I click the link you have there, the first post they have talked about peering. lol. So looks like they took that sort of thing into account.
Oh yeah they completely covered it and how to overcome it so it doesnt seem like much of an obstacle at all
Pretty cool that its community driven
Hmm, to some extent, certain assumptions can be achieved.
A lot depends on where it starts.
I’ll tell a little story.
In a previous life I was a bit involved in Amateur Computer Networks and later in running a very small ISP. Later, however, I watched the evolution of some networks.
I have also seen the rise and fall as well as the transformations of the little ones IX.
The model that has survived the years against the background of various others is consolidation. Small ISPs and very small ix united under one legal banner and started buying volume as one.
Started buying dark fibers as well as leased links in DWDM/CWDM tracts from suppliers in the area, through one central point in major DC they started buying from Tier1 like Level3, Telia and others.
This meant that even a small local ISP was able to have a very good price per Mb in good quality. And this allowed some to survive in confrontation with large corporate ISPs who often take individual clients for worms.
Building and maintaining a small IX and connecting to it has a lot of limitations for local ISPs due to the type of traffic that clients generate.
You can have 100 participants from IX but the local small ones do not generate any significant traffic volume among themselves. Transit via T2/T1 upstream and not local peering is important.
The big ASs will never agree to an exchange whether it’s a direct link or through some IX.
There are quite a few IX’s in the world with big Tier 1 presences, but just because you connect to ix won’t give you that route to them via ix. You still have to push traffic through the upstream.
In matters of IX and in general various concepts, nothing particularly new was invented. The problem is the cost for small networks.
I saw a large number of small IXs that were created with great joy, and died very quickly because they did not make much sense, unfortunately.
At the end of the day, it’s all about money. And no one will spend money if it’s pointless.
IX exist and will exist, new ones will also arise and others will disappear. In general, only those who have a sense of existence and thus rationalize the fact of spending money will survive.
The problem of scale, the smaller the network and the greater the distances, the worse it is financially calculated.
We have to ask ourselves what we really want to achieve.
We want to create a purely ideological IX or something that actually has a technical and financial logic behind it.
My experience shows that the consolidation of small entities and acting as a group allows you to achieve your goals.
Interesting story. I guess im confused on the takeaway
Are you saying because they grouped up as can be seen here.
That the financial feasability and survivability is greater than a few people running an IX?
Why not a mix of both
Im tired of the big corp ran internet but also im tired of how functionality and reliability and financial reasons are always put forth past security and the first principles of what the internet was supposed to be
Not a bad article on it
I am speaking in general terms and not necessarily about this particular one.
Building IX that started with a similar model I’ve seen many times.
This is especially more known and common in Europe than in the US, where more things have a corporate sticker.
Generally speaking, if I’m starting an IX, it must make sense. The sense of traffic exchange between the participants. However, the question is always the same whether the traffic that will be exchanged is worth the $ spent.
Here is an example of a few entities who were already in DC and decided to get together, I’ve seen such situations a few times and they are quite normal, that’s how many IXs were created in Europe. This is how one large IX was created in Central Europe, which was later bought and today exists under the banner of a large corporation.
In the USA, perhaps the model of creating IXs in the form of a loose group of people who start doing something on word may be a little less popular, but from my experience it was/is quite normal in Europe where many IXs were created as a small grouping of several companies/people, and if They hit the “sensible” point, they started to grow. Not every IX was created by some big ISP/DC.
Ahh thanks for the clarification. I think it made sense for these individuals. I mean at some point if you all are connecting to each others stuff. All visit common pages (the one i mentioned setup a google global cache server)… It might make sense at some point. All that said their primary driving reason seemed to be that most of them were network engineers and did it for fun
I wonder what the “makes sense” threshold is
Many IX’s are based on need, the problem is money unfortunately, you would like to separate it but it’s impossible, and it’s not about earning apsket but about the cost of existence itself.
If you are present with your toys in some DC and there are others there too then connecting within IX is not that expensive. Although almost every DC charges for every fiber from rack to rack, especially if the racks of the people involved are not next to each other. but these are still not terrible costs… The problem begins when you want to join this IX but you don’t have your toys in DC, here you have to rent fiber/bandwidth in tract systems from someone on the market to connect your location with DC, then it’s not cheap at all.
And here begins the adventure with which I had some experience, the consolidation of many small entities under one large one and a better position to negotiate the price for both a-b fiber and the capacity for transit from Tier1.
As I mentioned, a lot depends on where we start… The problem is not big when everyone is present in one place and either have their entire infrastructure there or a fiber plugged in for their location.
However, the costs start when we want to get to one place from many distant places, then the costs grow quickly and the question of the sense of such an IX begins to pose.
Construction IX when everyone is already in place and bear the costs of this presence as part of another need. Then crafting IX is just an add-on that doesn’t cost much.
However, the situation changes extremely when we have to get from many places to the central place where we start building IX. The costs start to be high and if we generate them only because of IX, the question must arise whether it makes sense. Ideologically yes, but financially often no.
Let’s say we’re building L1IX tomorrow. 
Let’s assume that the collection site will be located in the Equinix DE1 in Denver.
Now me as TH01 and you as PLL01 and Wendell01 from our locations we need to get transmission to Equinix DE1.
We buy a service from someone on the market, for example from Level3, who will set it up for each of us, let’s say 10Gb of interface between each of our locations and DC.
At Equinix, we rent a rack for our colo and pay to run three L3 fibers into our rack.
What costs will we pay for the existence of such an L1IX? Quite a lot… It would be different if our businesses already paid for the presence in DC because it is required by the company’s operation and we would only add IX as an addition.
That makes a lot of sense given (i think your in canada), im in SLC on a central fiber node and wendell is… Lets just say generically east coast for his privacy.
Thats LOTs of fiber backbone to rent. We arent colocated and we arent close so I imagine the distance plays a big role.
Now lets say you had three people on the forum start a few smaller IXes in denver, SLC and Socal with their buddies in the area. Those smaller IXes could probably share the backbone costs togetger wilst remaining community driven correct?
I mean ultimately that might mean setting up SLC as a location for the main IEX in this case since it would be centrally located
However if I understand you properly your basically saying that these would have to start locally with a reason for all those folks to peer and would only expand regionally IF those various smaller IXs needed to peer and it financially made sense to do so?
Also was DC meaning the capital or did you mean something else?
As a summary… Ideologically it makes sense and making IX is fun but we can’t pretend that money doesn’t matter here. 
I would like the internet to be free and everyone to have 10g in any location on the planet.
Whenever I use DC I mean Data Center, I always write cities as a whole.
Theoretical discussion is problematic because many different distances and financial reasonableness can be assumed for different people.
Everything is possible, but someone has to pay for it.
It’s hard to convince someone to pay if it’s of marginal importance to such a person.
That’s why I break it down into two aspects, ideological and financial.
As for the aspect of how much it has to be local or not, that’s also a big topic. IX can be as local as one building or per country or continent in terms of how far away its participants are. Look at big IXs like AMS-IX, DE-CIX, LINX and many more. They are located per one or several locations, usually within one city, creating one network IX. Participants come from all over the geographical area. Some are more local and others have their main DCs located in other countries.
The topology depends on the particular model and the situation where who is located. Sometimes it may be cheaper to put together a link between A-B than to pay for a colo… There are many combinations and you can complicate them depending on what will cost cheaper in a particular case.
I know situations where small ISPs help themselves with the transit of traffic between A-B. But these are very individual cases and it’s hard to talk about them in a general tone.
Generally, if money is not an issue then there are no limits when it comes to link setup. You don’t have to buy dark fibers or build your own transmission cables right away. This is why data transmission services and tunnels are sold, based on the transmission networks of various providers offered on the market. You can buy a tunnel carried in the IP layer, or a virtual fiber based on the optical carrier networks of the service provider. There are a lot of possibilities, it’s a matter of having $
Yeah i get that. Firstly I do have access to RESIDENTIAL 25 Gigabit symmetric connections through a few fiber providers here. (Perks of a very large centrally connected city)
So the connection itself to my home wouldnt cost more than the 210.99 a month it costs me to purchase that. That said the expense gets into bigger numbers when we talk interconnects and backbone
Pretty sure (but dont quote me)… Thats how hurricane electric works right? They are a broker of such things
We can also ask @wendell, because he probably leases some fiber, if I understand correctly, for off-site backup in a fairly close distance between A-B.
Yes, they sell a lot of it including colo and still seem to sell IP transit too.
This is neat, thanks for sharing!
You two are arguing about IX and DCs and meanwhile I am getting envious here in the middle of the ocean nowhere wondering “laying down fiber?”
Im like “to where?”
This will cost me an arm and a leg and arms and legs of other people and their extended family’s family…
So for me its like I’ll lay down my own indepent fiber under the sea or piggy back off data centers…
So why can’t we start as a small ISP, then build up the cash and meet people to get all the peering in place.
I think Tim summarized it in his prior post. They eventual get bought by ISPs, because operating costs are a thing.
