The addition of guarded fields

 bonito at nis.garr.it (Antonio_Blasco Bonito) writes:
  * I think we, as part of the european networking community, should pursue
  * the spreading of technical knowledge among network providers and
  * network operators. IP routing is not a trivial issue for the time being,
  * and the best results in routing coordination is obtained, in my opinion,
  * when the largest possible number of network operators do the right thing.
  * This is the spirit of the Internet, I think.
  * The NCC is a neutral organizations but this does not mean they should
  * act on behalf of network operators. The net result could be a minor
  * diffusion of relevant technical knowledge.
  * Service providers and even large network users communities should know
  * what they are doing when registering networks, creating guarded fields.
  * 
  * Also, local registries (down the tree from the NCC) must improve their
  * efficiency. The Internet is growing very very fast in Europe and we
  * have to use solutions that scale. A do-all NCC is not the solution, I guess
  * .
  * I think the NCC is going is the opposite direction (i.e. proposing
  * the PRIDE project, which I feel a very important activity in the next futur
  * e)
  * 
Thought I'd comment on this a little as I'm currently working on much
stronger syntax checking for the new databsse software. Whilst it is not
complete it's pretty careful and it raises exactly the point Blasco brings
up. Here are the basic results on the Database breaking them down by object.

 Object   Errors    Oks   Warnings     Total    (% in error)

aut-num     47      47        0           94         50.0
bdry-gw      0       7        0            7          0.0
domain    1284     913        0         2197         58.4
inetnum    899    8566        0         9465          9.5
person    1087    8824       10         9921         11.0
rout-pr      0       3        0            3          0.0

This is pretty bad as you can see. This includes much stronger checking of
mandatory fields and syntax as they are written in the documents which is why
the aut-num is high as we currently have a lot of missing mandatory fields.
Once the new software is in place the syntax checker will spot all of 
these. However, it still remains that many entries are submitted without even
checking the syntax of the object.  At some point we must also remove
these errors although this needs some careful thinking about. Any ideas ?

Of course one can argue that stronger syntax checking should have been done
from day one but it is historic as part of this original software was done 
before their was an NCC. However, please can we all try to be careful 
with updates at least til' the software is fully in place ;-).

--Tony.