Re: [lisp] LISP EID Block Size

[Paul Vinciguerra <[e-mail]>]

I think this issue here is that the request is for a solution without a problem.

If you want to do experiments, lobby cisco to create a configuration option of a list of, say "known-EID 153.16.224.0/20".    

That can be configured for these experiments.

I have unused EID space, I am sure Sander and Cisco do as well. I have offered address space since we met back  in Orlando.

Once there is a track record and a real need for a block, then we have that much more evidence to support a future request.  If the experiments can be done this way, then the block wasn't actually needed in the first place.

But the way this is positioned, "LISP" wants an address block.  I am not sure what that means. Is it the working group? Is it Cisco? Who is going to be the gatekeeper? Who will decide what experiments have or do not have merit to use part of the block.

To float a real life LISP example:

We are part of another LISP experiment, the DDT.  DDT is a great technology enhancement, not for the database indexing aspect, but for map-server peering that allows for a more resilient topology.  I am all for experimentation.  

The public DDT roots experiment is a great example of such gating.  Cisco is the gatekeeper here.  Their DDT clients only allow for 4 roots per AFI.  And its filled by the gang of four: Cisco, InTouch, Us (Vinci Consulting/VXNet), and Verisign.  I was there when Mike Gibbs identified this over a year and a half ago and was assured it would be expanded.  It's still the same today.  Whether this is intentional control being influenced by the vendor, or just a lack of resources, the outcome to the end user is the same.  Depending on your relationship to the vendor, you can choose your position as to the reasons why.  

So I come back to my original question:  Why do you need this block?

Is is that you don't want to "rent" addresses?  
I would submit that this debate has cost all our employers more than any block would cost for a decade.  

Is it that the group doesn't want to renumber in the future?  
State it.  I think the group gets credibility for admitting it.

Is it going to minimize changes to the code?
Any experiment _is_ a code change, no?

Is the performance of a single large prefix orders of magnitude better that a handful of configured prefixes?  
You all would know that much better than I would.  

I personally think that more will get done with small agile groups that would be empowered to try something on their own and succeed or fail quickly, rather than this large command/control structure that requires merit to be proven before the experiment to show the merit of the experiment being tried.

Paul

________________________________________
From: [e-mail] [[e-mail]] on behalf of Luigi Iannone [[e-mail]]
Sent: Sunday, November 03, 2013 5:30 PM
To: Dino Farinacci
Cc: LISP mailing list list
Subject: Re: [lisp] LISP EID Block Size

Hi,

I agree with Dino,

if the issue is just the size let’s reduce it and do some experiments.

On the other hand, I do not understand we people here are trying to reach a binary conclusion like “EID Block is important and useful” or “EID Block is useless” even before doing any experimentation.

IMHO this is not the most logical order. We should first experiment, then we will have to know-how to make a decision.

Exactly because there are different and opposite opinions let’s the technology itself, through experimentation, make the decision.

Luigi