[Nsi-wg] Topology section

[John MacAuley]

I didn't literally mean the LDAP protocol itself, although, I would definitely argue the "it does not scale" argument.  I was trying to relate a simple interface similar to LDAP (which is a stripped down version of X.500) that would allow me to query a peer's view of the topological world.  In addition, through simple notifications I could be told when changes have occurred on topologies of interest.

I like Jeroen's P2P angle.  I did a lot of work with the peer discovery mechanisms back when bit torrent first came out as a way to reduce tracker load.  The concepts of nodes and super nodes to distribute topology is a cool idea, but we have a restricted peering model that would prevent any random NSA from communicating with any other random NSA.  Of course, we could decide to change our model to more of a P2P relationship and what the sysadmin's across the world have heart attacks ;-)

John

On 2012-05-30, at 6:42 AM, Radek Krzywania wrote:

> Hi,
> Regarding LDAP - it does not scale. It's just simple tree structure, not a graph so we can’t model too much with that. Never heard of any mechanisms for distributed maintenance. IMHO - Pro: easy to implement, Cons: all the rest.
> 
> Best regards
> Radek
> 
> ___________________________________
> Radoslaw Krzywania          
> 
> Network Research and Development
>  Poznan Supercomputing and  
>      Networking Center
> 
> radek.krzywania at man.poznan.pl
> +48 61 850 25 26             
> 
> http://www.man.poznan.pl
> ___________________________________
> 
>> -----Original Message-----
>> From: nsi-wg-bounces at ogf.org [mailto:nsi-wg-bounces at ogf.org] On Behalf
>> Of Jeroen van der Ham
>> Sent: Wednesday, May 30, 2012 11:59 AM
>> To: John MacAuley
>> Cc: NSI WG
>> Subject: Re: [Nsi-wg] Topology section
>> 
>> Hi,
>> 
>> On 29 May 2012, at 16:22, John MacAuley wrote:
>> 
>>> Hot dang, a heated debate.  I thought everyone had fallen into a volcano
>> while in Iceland.
>> 
>> Some of the fire from the volcanoes spurred us back to the debate indeed ;)
>> 
>>> I nearly swallowed my tongue when I read OSPF.  I was hoping for
>> something extremely simple that would just allow me to query a peer and
>> control the retrieval of what they know. Something very similar in concept to
>> a protocol like LDAP where I can list the top level branches of the tree
>> (available networks), then do a detailed retrieval of the contents of a subtree
>> (topology for the network).  I would also like to put a watcher on a subtree to
>> be notified when anything was updated.
>> 
>> I have no close experience with LDAP, how does it work with multiple
>> distributed sources of information? What about the subtree notifications?
>> 
>>> I am definitely big on reuse, but if my aging memory serves me correctly,
>> the last time I implemented OSPF in a product it was not a trivial task.  I need
>> a bit more of trivial these days ;-)
>> 
>> I indeed meant an OSPF-like protocol.
>> It may not be trivial, but it's a proven technology. It has some great
>> extensibility features using the TLV fields.
>> 
>> If that's off the table, we could of course also look into peer-to-peer like
>> systems. There is some great work on distributed storage using distributed
>> hash tables (DHT) that may also be applicable to this situation.
>> 
>> Jeroen.
>> 
>> _______________________________________________
>> nsi-wg mailing list
>> nsi-wg at ogf.org
>> https://www.ogf.org/mailman/listinfo/nsi-wg
>