How about this:

I'm happy that you are writing up all of this info, but maybe find your own page to do it? I don't care if you didn't like my comments. If you want to address where I was wrong, then do it in the page I started. Just deleting some bodies thoughts is the nature of wiki, but in my mind it would be proper netiquette to provide inline comments.


I didn't really want to get into ThreadMode on this.

EJ> But if you see my first post about this, I was asking for inline comments.

The assumptions that you made in your original page were based on a misunderstanding of BSS and IBSS. They were fundamentally wrong.

EJ> You still haven't shown me why.

EJ> First off, we are not connecting between LANS, we are creating a MAN. fundamentally, they are all just network interfaces, with different layer 2 protocols/handshakes. So I connect Node1 and Node2 via smoke signals.

EJ> But my wiki page didn't talk about BSS Roaming. I was talking about Point to point links. I talked about Roaming in your reply about BSS clients talking to more than one BSS masters.

EJ> OK, fine. if we are only using 802.11b as a transport, what does it matter what mode we are in? If you want to do layer 3 routing, as your PointToPoint page talks about, WHY does it matter?

In addition to staying vendor neutral, we must keep costs down or there is no way we will be able to get anyone to put up nodes. Using the scheme that you laid out requires massive outlay of capital from every user for no reason other than 'the gear is there already'.

EJ> I'm all for staying vendor neutral. But I believe the solution exists, and we are working on things that don't fit in the model you are describing. If we are protocol nutral, and vendor nutral, why does it matter what hacks we use to make it work? As for the outlay of capital, how much does a pair of WAP11 cost, with antennas & sealed boxes? $400 a node? By the time you purchase the SBC & WLAN cards, the cost will be the same.

This project is not only about building a network, but designing a network that works well and will scale to fit our needs.

EJ> Then let's design the network, stick with layer 3 routing, and leave the layer 2 stuff up to any vendors we want (being vendor nutrual & all).

I understand that I have been saying IBSS is the answer without qualifying why. This has been a mistake on my part, and something that I am now trying to correct.

EJ> My proposal was the inverse, with the same end result in mind: "Why not this?"

I was under the assumption that people on the list understood the limitations of 802.11b and what we were trying to accomplish.

I am now trying to explain WHY without pointing people to an IEEE document and saying RTFM.

I did not feel that I could respond to the points in your original page without us both having an understanding of the underlying technology, what the differences are between layer 2 and layer 3, and the goal that we are trying to accomplish.

EJ> But as you stated above, we don't care about the underlaying tech, vendors, and layer 2 stuff. So what does it matter?

I am sorry that I deleted your original page, but it adds confusion to the issue.

EJ> Thank you for the apology.

The issue here is education and hopefully I can correct that.

By the way, I am not suggesting that ClientNodes connect to IBSS links. BSS has it's place, but it is at the end of the network, not the core.

The document that I am writing is now at PointToPoint

EJ> It's looking great! Would you mind if I add some inline questions?

PtpMode (last edited 2008-04-13 16:36:14 by localhost)